Coder Legion - Recent posts tagged aeo

Two thirds of what Google's AI Overviews cite comes from the top 30% of a page.

Tue, 29 Sep 2026 23:46:22 +0000

We checked 600 real searches and traced about 4,300 cited passages back to the exact spot on the page they came from.

67% came from the top 30% of the page in Google AI Overviews, and 60% in Google AI Mode.

Kevin Indig's research found the same pattern for ChatGPT, but weaker: 44% of its citations come from the top 30%.

Takeaway: don't bury the lead. Put the answers you want AI to cite in the top 30% of your pages.

We're repeating these searches on a regular basis to see how much this changes. Follow along for the results and more of what we're finding about how to get cited by AI.

We ran an AI SEO audit on 50 SaaS sites, here are the key findings

Wed, 23 Sep 2026 22:56:23 +0000

36% are never read. This was the biggest issue we found and it was due to their pages being empty until JavaScript runs. With the exception of Gemini and AppleBot, no AI crawlers execute JS so they get a blank shell on every request.

53% of sections are under 40 words, and the median section is 35 words. Retrieval hands an assistant one passage at a time, usually 100 to 300 words. This means 53% of sections are too short. They may still get quoted, but without enough information in the answer they are set up to lose.

Only 28% of sections contain a number, a quotation or a link. Based on a 2024 GEO study, AI likes numbers, something to quote or a link so this could lead to not getting a mention. It's a particular worry for smaller businesses with lower search page rankings.

From the study I mentioned: adding statistics, quotations and citations were the three best-performing of nine tested content changes, up to 41% better than baseline. Specifics get pulled into the answer; adjectives get dropped.

60% bury the answer below the first 200 words. This can lead to AI not getting the answer in the chunk it lifted and you missing out on a mention.

92% keep content on subdomains their main robots.txt does not cover. Another tip, ensure your blogs, docs or anything else on a subdomain has its own robots.txt.

If you want to run a free audit on your site here's the link: https://babel42.io/ai-seo-audit

From SEO to AI Search: Jarno S’s Journey Into Making the Web More Understandable to AI

Fri, 18 Sep 2026 10:22:19 +0000

From SEO to AI Search: Jarno S’s Journey Into Making the Web More Understandable to AI

“Build for people. Structure for understanding. Measure what changes.”

Search is changing.

For years, businesses and website owners have focused on questions like: Where do we rank on Google? How much traffic are we getting? Which keywords are bringing visitors?

But as people increasingly turn to AI systems such as ChatGPT, Gemini, Perplexity and Copilot for answers, another question is becoming increasingly important:

What does AI say about your business — and is it accurate?

For Jarno S, founder of AEOvara in Finland, this question sits at the intersection of technical SEO, AI search, structured data, content architecture and measurement.

Jarno's path into this work has been practical rather than purely theoretical. He has built websites, worked with visual content and photography through Kuvaajankulma, and experienced how difficult it can be for a good business to become discoverable online.

SEO became a natural part of solving that problem.

Then AI search changed the question.

Instead of asking only whether a website can rank, Jarno became interested in whether AI systems could understand a business correctly, represent it accurately and recommend it for the right reasons.

That thinking led to AEOvara and his work around Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), and Large Language Model Optimization (LLMO).

From Traditional Search to AI Search

Jarno describes several related concepts that are increasingly appearing in conversations about search.

Traditional SEO is mainly concerned with earning visibility in search-result pages.

AEO — Answer Engine Optimization — focuses on structuring content so that it can answer a question directly.

GEO — Generative Engine Optimization — focuses on how a brand is represented and recommended in generative search experiences.

And LLMO — Large Language Model Optimization — is about making a website, its entities and its information understandable to large language models.

These approaches overlap, but for Jarno the broader shift is more important than the terminology.

The goal moves from simply trying to “rank for a keyword” toward becoming a reliable answer.

That distinction becomes particularly important when the interface between a person and the web is no longer simply a list of links.

The Question That Changed Everything

Jarno became interested in AI search when he noticed that people were increasingly asking ChatGPT, Gemini and Perplexity questions that they previously asked Google.

That introduced a new problem.

It was no longer enough to ask:

“Where do we rank?”

There was another question:

“What does the AI say about us, and is it accurate?”

That question became central to his work.

Jarno describes his work as having two audiences: the people reading a website and the AI systems trying to understand it.

That idea has influenced how he approaches website architecture and content.

A page should work for a person who is in a hurry, while also giving a system enough clarity to identify the business, its services, evidence and expertise.

That means clear page purpose, direct answers, logical headings, consistent facts and strong internal connections.

AI Visibility Should Be Measured

One of Jarno's strongest themes is measurement.

AI search can sometimes appear mysterious. Ask an AI system a question, receive an answer, take a screenshot and it can be tempting to conclude that a business is either visible or invisible.

Jarno argues that this isn't enough.

AI outputs can vary depending on the platform, wording, location and time.

A more useful approach is to create a repeatable measurement process.

His recommendation is straightforward:

Start with 8–15 real customer questions.

Keep the wording stable.

Test those questions across the relevant AI platforms.

Then record the results.

The goal isn't to create a perfect score.

The goal is to establish a trustworthy baseline that makes changes visible.

Think of AI Visibility Like Regression Testing

Jarno compares an AI visibility benchmark to a software regression test suite.

The analogy is particularly useful for developers.

If the test changes every time you run it, it becomes difficult to determine whether the software changed or the test changed.

The same applies to AI search.

If the prompt changes every time, you cannot reliably determine whether a change in the result came from an improvement in the brand's visibility or simply from asking a different question.

Jarno recommends keeping a stable core set of prompts while adding new questions over time when necessary.

It turns AI visibility from a collection of interesting screenshots into something that can be observed and compared.

A Mention Isn't the Same as a Citation

Another distinction Jarno considers important is the difference between mentions and citations.

An AI system might mention a business in an answer.

But that mention could be vague, incorrect or simply part of a long list.

A citation provides a stronger signal because it connects a specific claim to a source.

Both matter, but they answer different questions.

Visibility is useful.

But trusted and accurate attribution is more valuable.

For Jarno, metrics should therefore be separated rather than combined into one simplistic number.

He focuses on areas such as:

Visibility
Mentions
Citations
Accuracy
Meaningful or qualified actions

The distinction helps businesses understand not only whether AI systems are talking about them, but how they are representing them.

Structured Data Helps AI Understand the Pieces

Jarno's work also places significant emphasis on Schema.org structured data and entity clarity.

He doesn't describe Schema.org as a magic solution.

Instead, he sees it as a useful shared vocabulary that can help systems identify things such as:

Products
Services
People
Organisations
Locations
FAQs
Relationships between entities

But structured data becomes much more useful when it matches the visible content on a website and the information remains consistent across the site.

Adding schema isn't a substitute for having clear information.

The underlying content still matters.

What About `llms.txt`?

There has been considerable discussion around llms.txt as websites adapt to AI systems.

Jarno takes a measured approach.

He sees llms.txt as an experiment and a helpful orientation layer, rather than a visibility button.

It may make important pages and topics easier to discover, but it cannot replace useful content, technical quality, reputation or independent evidence.

Jarno's approach is simple:

Implement sensible experiments, then measure whether they actually change anything.

That philosophy is consistent throughout his work.

Don't adopt something simply because it is currently being discussed.

Test it.

Measure it.

Then decide what the evidence tells you.

Make Pages Easy to Quote

One of Jarno's ideas is that websites should be “easy to quote, not merely easy to crawl.”

That requires thinking about how information is presented.

His practical recommendations include:

Put the main answer where people can find it quickly.
Use descriptive headings.
Explain important terms clearly.
Support claims with evidence.
Show authorship.
Include update dates where appropriate.
Avoid hiding important facts inside marketing language.

The goal is to make information easier for an AI system to understand and represent accurately.

But these practices also make websites easier for humans to understand.

That overlap is important.

The AI Search Advice He Thinks Is Overhyped

The AI search industry is moving quickly, and with that comes a constant stream of new tactics.

Jarno is skeptical of the idea that one file, one prompt trick or one piece of schema can suddenly make a business “rank in ChatGPT.”

For him, AI visibility isn't a switch.

It is built from several things working together:

Helpful content + technical accessibility + entity clarity + credible references + ongoing measurement.

This is also why he prefers experimentation over blindly following trends.

If something has a plausible mechanism, can affect how people discover businesses and can be tested responsibly, he sees value in running a small experiment.

If it creates excitement but cannot be measured or explained, he treats it as noise until stronger evidence appears.

What AI Search Reveals About Businesses

One of Jarno's biggest surprises has been how uneven AI search results can be.

A business may be described accurately on one platform and poorly — or not at all — on another.

Even more interestingly, AI systems can sometimes produce confident-sounding answers when the underlying business information is incomplete or mixed up.

That creates a different kind of challenge for businesses.

It's not simply about gaining visibility.

It's about making sure the information that becomes visible is correct, consistent and supported by evidence.

Building AEOvara

Jarno founded AEOvara to help businesses navigate search in the AI era.

The challenge is partly technical and partly educational.

Many business owners already understand concepts such as Google rankings and website traffic.

But asking:

“How do AI systems describe my business?”

is still a relatively new question.

Jarno believes the answer is to keep the conversation practical.

Instead of selling mystery, show businesses:

What AI systems currently say
Where information is missing
Where information is inaccurate
What can be improved
What changed after the work

That makes a new category easier to understand.

It also keeps the focus on measurable outcomes rather than hype.

From AI Visibility to Repeatable Benchmarks

A major part of Jarno's current exploration is making AI visibility more reproducible.

He has developed an ongoing AI visibility benchmark and published practical guides around measuring how AI systems discover and cite content.

The idea is straightforward: if businesses want to improve their AI visibility, they need a way to observe what is happening before and after making changes.

That means stable prompts, repeated observations and clearly separated metrics.

Instead of asking whether a website is “good at AI search,” businesses can begin asking more specific questions:

What changed?

Where did it change?

Was the information accurate?

Was the business cited?

Did the change lead to a meaningful outcome?

Those questions turn an abstract concept into something that can be investigated.

Lessons From Photography

Jarno's work isn't limited to SEO and AI search.

He also has a background in professional photography and visual content through Kuvaajankulma.

That experience has influenced how he thinks about discoverability and trust.

Photography taught him that presentation matters.

A website needs a clear point of view, coherent presentation and real evidence.

It also taught him to respect the user's attention.

The best page isn't necessarily the one containing the most information.

It's the one that makes the right information easy to understand.

That principle carries naturally into his thinking about AI search.

Where Search Goes Next

Jarno doesn't see traditional search disappearing completely.

He expects it to remain important, particularly for research, navigation and comparison.

At the same time, he believes AI assistants will increasingly become the starting point for many questions, especially broad or conversational ones.

Rather than imagining a complete replacement, he sees a more blended search journey emerging.

People may move between traditional search engines, AI assistants, websites and other interfaces depending on what they are trying to accomplish.

For businesses, that means the underlying information needs to work across more than one type of discovery experience.

Advice for Developers and Small Businesses

For developers, technical founders and small businesses thinking about AI search, Jarno recommends starting with fundamentals rather than chasing every new trend.

His practical starting points are:

Keep business facts accurate and consistent.
Make important pages technically accessible.
Answer real customer questions clearly.
Use relevant structured data.
Make it clear who is responsible for the content.
Build a stable set of prompts and measure them over time.
Focus on changes that can actually be verified.

The common thread is simple:

Improve what you can measure.

A Lesson From Building a New Category

One lesson Jarno learned while building AEOvara is that too much jargon can make a new field harder to understand.

Technical terminology may be useful internally, but customers ultimately need answers to practical questions.

Are we being understood correctly?

Are we being recommended for the right reasons?

What changed after the work?

Those questions provide a much clearer way to explain the value of AI search visibility.

AI Search Is an Information-Quality Challenge

Jarno believes one of the biggest misunderstandings around AI search is treating it simply as a temporary traffic-generation trick.

He sees something broader.

AI search is also an information-quality challenge.

If a business cannot explain itself clearly, consistently and with evidence, AI systems will have a harder time understanding it correctly.

That makes AI visibility partly a technical problem, but also a content, information architecture and trust problem.

What's Next for AEOvara?

Jarno is most excited about making AI visibility more reproducible.

That includes practical benchmark methods, clearer reporting and tools that help smaller businesses understand how answer engines see them.

AEOvara has already explored this through practical guides, an llms.txt generator and ongoing AI visibility benchmarking.

The broader goal is to make AI search visibility less mysterious and more measurable.

Rather than asking businesses to trust a new acronym or a new optimization trick, the approach is to establish a baseline, run experiments, observe changes and learn from the evidence.

That mindset — measure what changes rather than simply following the latest trend — runs through Jarno's approach to AI search.

About Jarno S

Jarno S is the founder of AEOvara, based in Lappeenranta, Finland.

His work focuses on Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), AI Search Optimization, Technical SEO, Schema.org structured data and LLM Optimization (LLMO).

He works across search ecosystems including Google Search, ChatGPT, Gemini, Perplexity and Copilot, exploring how AI answer engines discover, interpret and cite web content.

His current work includes entity clarity, structured data, repeatable AI visibility benchmarks and practical ways to connect traditional SEO with AEO, GEO and LLMO.

His guiding principle is simple:

“Build for people. Structure for understanding. Measure what changes.”

CoderLegion Developer Stories

This story is part of CoderLegion Developer Stories, where developers, engineers, technical founders and technology professionals share their experiences, challenges, projects and lessons from their journeys.

The goal is to give the CoderLegion community a closer look at the people and ideas behind the technology.

Interviewed and edited by Mehadi Hasan, Community & Editorial Team at CoderLegion.

Your AI Visibility Score May Be Wrong: A Reproducible Testing Protocol

Fri, 11 Sep 2026 20:10:27 +0000

By Jarno Saarimies, Founder of AEOvara. This is Part 3 of the series How to Build a Repeatable AI Search Visibility Benchmark.

Ask ChatGPT the same commercial question twice and you may get two different shortlists. Change “best” to “most reliable,” and the shortlist may change again. Run the prompt in a continuing conversation, in another country, or after a model update, and you may no longer be measuring the same system.

That creates an awkward problem for AI visibility reporting: a dashboard can show a precise score even when the experiment behind it is not precise.

This does not make AI visibility measurement useless. It means the score must be treated as an estimate produced by a documented test—not as an objective ranking retrieved from a fixed index.

In Parts 1 and 2 of this series, I explained how to build a prompt corpus and track mentions, citations, competitors, and answer quality. This article addresses the harder question:

When is a change in AI visibility evidence, and when is it only measurement noise?

The answer requires more than repeating prompts. You need to control the test environment, preserve the raw evidence, quantify uncertainty, and separate effects that dashboards often collapse into one number.

First, identify what actually changed

Suppose your mention rate rises from 30% to 40%. At least three different things could have produced that result.

1. The brand changed

You published new evidence, earned independent coverage, corrected an entity description, improved a relevant page, or became more widely discussed. This is the effect an AEO or GEO programme is trying to measure.

2. The engine changed

The provider released a new model snapshot, changed retrieval behaviour, refreshed an index, altered citation rules, or modified the product. OpenAI explicitly states that model outputs are variable and that prompting behaviour can change between model snapshots. For API applications, it recommends pinned model versions and evals when consistency matters.

Consumer AI products rarely give researchers complete control over those variables. A visible product name such as “ChatGPT” is not a permanent experimental condition.

3. The sample changed

The wording, session history, location, language, login state, retrieval mode, time, or number of repetitions changed. Even if the website and model stayed untouched, the measured score could move.

If your log cannot distinguish these explanations, it cannot support a causal claim such as “our optimization increased AI visibility by 10%.” It can only say that the observed answers changed.

That distinction sounds cautious. It is also what makes the result credible.

A prompt is not a keyword

Traditional rank tracking usually observes a defined query, location, device, and search engine. AI prompts carry more semantic degrees of freedom.

Consider these questions:

What are the best technical SEO consultancies for a Finnish SME?
Which technical SEO consultancy is most reliable for a small Finnish company?
Who can help a Finnish business become visible in AI-generated answers?
Recommend an affordable Finnish specialist in SEO and AI search optimization.

They occupy the same commercial neighbourhood, but they are not interchangeable samples. “Best,” “reliable,” “affordable,” and “AI search optimization” introduce different selection criteria.

A 2026 controlled study by Jan Ehrlinspiel, Malte Landwehr, and Tomek Rudzki examined 37,804 responses across five AI answer engines. It found that prompt format and archetype can shift visibility baselines, with the clearest sensitivity appearing in unbranded, middle-funnel commercial prompts. The authors are affiliated with Peec AI, so that commercial relationship should be disclosed, but the methodological lesson is useful: one wording cannot represent an entire buyer intent.

Freezing a prompt list is therefore necessary, but not sufficient. A frozen list gives you repeatability for those exact prompts. It does not prove that the list represents the full market.

Use prompt families, not a bag of convenient questions

A stronger benchmark separates two layers:

Core prompts remain unchanged so trends can be compared over time.
Prompt variants test whether the result survives realistic changes in wording.

For each buyer intent, write one core prompt and two or three variants before checking which version favours your brand. This prevents an easy form of unconscious cherry-picking: keeping the wording that produces the most flattering result.

Buyer intent	Core prompt	Controlled variants
Category discovery	What are the best AEO agencies in Finland?	most reliable; established; suitable for SMEs
Problem solving	How can a Finnish SME improve visibility in AI answers?	get cited by ChatGPT; appear in AI search; improve generative search visibility
Supplier selection	Who should I hire for an AI visibility audit in Finland?	independent specialist; consultancy; local expert

Do not average branded questions such as “What is AEOvara?” into the same score as unbranded discovery questions. A branded prompt measures recognition and description accuracy. An unbranded prompt measures discovery. Combining them can make a weak discovery result look healthy.

Record the test environment like an experiment

Every observation should include enough metadata for another person to understand what was tested:

exact prompt text;
platform and visible product mode;
model identifier, when exposed;
web search or grounding status, when exposed;
date, time, language, and country;
signed-in or signed-out state;
clean session or continuing conversation;
personalization or memory state, when known;
repetition number;
full answer and cited URLs;
scorer name or scoring-rule version.

“ChatGPT, September” is not adequate provenance. Neither is a cropped screenshot that hides the prompt, date, citations, and product mode.

API-based monitoring improves control, but it creates a different test. An API model with a pinned snapshot and explicit search configuration is not necessarily equivalent to the consumer interface used by customers. Report API observations as API observations. Do not quietly label them “what users see in ChatGPT.”

Measure a rate—and show its uncertainty

If a brand appears in 3 of 5 runs, the observed mention rate is 60%. That does not mean the true probability of a mention is known to be 60%.

With only five binary observations, a 95% Wilson confidence interval is approximately 23% to 88%. The interval is wide because the sample is small.

If the brand appears in 30 of 50 comparable runs, the observed rate is still 60%, but the corresponding interval narrows to approximately 46% to 72%. More observations have not changed the headline percentage; they have changed how much confidence we should place in it.

For a binary mention metric:

mention rate = mentions / comparable runs

Report the numerator and denominator—not just the percentage:

ChatGPT web search: 12/20 mentions, 60%
95% Wilson interval: approximately 39%–78%

The interval is not decoration. It helps prevent claims that the evidence cannot support.

If period A is 6/20 and period B is 8/20, the dashboard has moved from 30% to 40%. But with samples that small, the difference may be ordinary variation. Call it an observed increase, not a confirmed improvement.

Do not use one noise budget for every metric

Mention, citation, prominence, sentiment, and factual accuracy behave differently.

Mention is binary and relatively easy to score.
Citation requires a link or source reference and should be stored at URL level.
Prominence depends on an explicit rule: first named, top three, recommended, or merely included.
Sentiment is a classification judgement and is usually noisier.
Accuracy must be checked against a dated source of truth.

A 2026 production-data paper on GEO measurement reported that sentiment was 6.7 times noisier than mention in its dataset. That is one vendor’s observational result, not a universal constant, but it illustrates why a single blended “AI visibility score” can hide more than it reveals.

Publish the components separately. A brand can have a rising mention rate while its citation rate falls or its description becomes inaccurate. Those are materially different outcomes.

Establish a noise floor before claiming a win

Before changing the site, run the unchanged benchmark over several measurement windows. This creates a baseline for natural movement.

Then document the intervention:

page or entity changed;
exact publication time;
URLs affected;
intended prompt family;
no other known campaign, migration, or major PR event;
first date the changed page was observed or cited.

Continue measuring with the same core prompts and conditions. Compare post-change movement with pre-change variation.

This is still not a perfect causal experiment. AI engines can update during the same period, and third-party sources can change independently. But it is much stronger than editing ten pages and attributing the next dashboard increase to whichever tactic sounds most impressive.

Evidence	Defensible wording
One answer changed	“The brand appeared in this observation.”
Repeated rate increased	“Observed mention frequency increased in this prompt set.”
Increase exceeds baseline variation	“The change is larger than the noise previously observed in this test.”
Controlled intervention with comparison group	“The intervention is a plausible cause of the measured difference.”
Repeated controlled result across engines and periods	“The evidence supports a robust effect under the tested conditions.”

Avoid jumping from the first row to the last.

Build a holdout set to catch overfitting

If you optimize pages against the exact prompts used for reporting, you may improve performance on the test without improving broader buyer discovery.

Reserve 20–30% of your prompt families as a holdout set. Do not use those prompts to decide what to write or edit. Run them only at defined validation points.

If the core tracking prompts improve but the holdout prompts do not, you may have optimized to the benchmark rather than the market. This is the AI-search equivalent of teaching to the test.

The holdout set should remain commercially relevant. Random prompts do not create scientific rigour; they only create irrelevant data.

Publish the method, including its limitations

E-E-A-T is not a numeric score that can be added with an author box or a block of schema. For readers—and for systems trying to evaluate a source—trust is easier to establish when claims have visible provenance.

A credible public benchmark should state:

who ran the test;
what was measured;
the exact period and platforms;
how prompts were selected;
how many observations were collected;
how answers were scored;
whether the evaluator had a commercial interest;
what changed since the previous version;
what the data cannot prove.

That is the approach used in the continuously updated AEOvara AI Visibility Benchmark Suomi 2026: the platform-level results, prompt scope, cadence, and limitations are part of the evidence, not hidden behind a proprietary score.

Publishing an unfavourable or inconclusive result can strengthen the work. A benchmark that only produces success stories looks like marketing. A benchmark that preserves negative findings begins to look like a useful dataset.

A minimum viable reproducibility checklist

Before publishing the next percentage, ask:

Are branded and unbranded prompts reported separately?
Were prompts written before seeing which ones favoured the brand?
Are the exact wording and realistic variants preserved?
Were runs made in clean, comparable sessions?
Are platform, model, retrieval mode, location, and date recorded?
Does every percentage show its numerator and denominator?
Is uncertainty visible when the sample is small?
Are mentions, citations, prominence, sentiment, and accuracy separate?
Was a pre-change noise floor established?
Is there a holdout prompt set?
Are raw answers and cited URLs archived?
Are limitations and commercial relationships disclosed?

If several answers are “no,” the result may still be useful as an observation. It should not be presented as a stable market measurement.

The bottom line

AI visibility is measurable, but not as a single universal rank.

What you can measure is a defined brand’s behaviour across a documented sample of prompts, engines, configurations, places, and times. Repeat that protocol and you can build evidence. Hide those conditions behind a polished score and you create false precision.

The strongest AI-search report is not the one with the most decimal places. It is the one another practitioner could challenge, reproduce, and learn from.

That is also the kind of source worth citing.

FAQ

How many times should I run each AI-search prompt?

There is no universal number. Three to five repetitions can expose obvious variability and provide a directional baseline, but a small sample does not produce a precise probability. Report the raw count and an uncertainty interval, then increase repetitions for important or volatile prompt families.

Should I freeze my prompts or update them?

Do both in separate layers. Keep a stable core set for trend comparison, and maintain a versioned discovery set that reflects new buyer language, products, and competitors. Never silently replace old prompts and continue the same trend line.

Can I compare an API result with the consumer AI product?

Only with an explicit caveat. The API may use a different model snapshot, system configuration, search policy, location input, or citation interface. Treat it as a separate measurement surface unless equivalence has been demonstrated.

Is a higher mention rate proof that my content change worked?

No. It is evidence that the observed rate changed. A causal claim requires stronger controls: stable test conditions, a pre-change noise floor, documented timing, enough observations, and preferably a comparison or holdout group.

Does a citation mean the AI answer used my page correctly?

Not necessarily. Store and inspect the cited URL, the claim it appears to support, and whether the answer accurately represents the source. Citation presence and citation correctness are separate metrics.

Sources and further reading

OpenAI, API backwards compatibility and model variability.
Ehrlinspiel, Landwehr & Rudzki, Prompt Tracking Works, But Not as a One-Prompt Measurement System, 2026.
Blago, Generative Engine Optimization at Scale: Measuring Brand Visibility Across AI Search Engines, 2026. Treat the vendor-operated production dataset and its generalizability accordingly.
Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024.

About the author: Jarno Saarimies is the founder of AEOvara, a Finnish consultancy focused on technical SEO, Answer Engine Optimization, Generative Engine Optimization, and reproducible AI-search visibility measurement. His work emphasizes documented methods, source verification, and claims that remain useful after the hype cycle moves on.

AEO for Finnish SMEs: A Practical 2026 Guide to Getting Cited by AI

Thu, 10 Sep 2026 10:07:28 +0000

Search has changed. Quietly, rapidly, and permanently.

For years, businesses focused on climbing Google rankings. But in 2026, the game is different. Search engines, AI assistants, voice devices, and smart summaries increasingly deliver direct answers instead of lists of websites.

That means Finnish small and medium-sized enterprises must adapt—or risk becoming invisible.

This is where Answer Engine Optimization (AEO) comes in. AEO helps your business become the answer users see first. Not just another blue link. Not page seven. Not forgotten.

For Finnish SMEs, that shift is massive.

What is Answer Engine Optimization?

AEO stands for Answer Engine Optimization. It is the process of making your website, content, authority signals, and technical data easy for search engines and AI systems to understand, trust, and cite.

Traditional SEO asks:

How do we rank first?

AEO adds a sharper question:

How do we become the recommended answer?

The two disciplines work together. SEO helps people and crawlers discover your content. AEO helps answer engines confidently extract and reuse the right part of it.

Why Finnish SMEs need AEO in 2026

Classic SEO still matters, but rankings alone no longer guarantee attention. AI-powered search experiences increasingly summarize information before a user visits a website. They favour businesses that communicate with clarity, demonstrate expertise, and provide structured, verifiable information.

For an SME, this can translate into:

More appearances in AI summaries and answer boxes
Better-qualified leads with stronger buying intent
Improved visibility for local and city-based searches
Greater trust through transparent expertise, reviews, and business information
Less dependence on paid advertising

In plain English: when someone asks an assistant for a provider, price range, comparison, or next step, you want your business to be easy to recommend.

The shift from rankings to answers

Being number one is useful. Being cited may be even more valuable.

A page can rank well and still fail to answer the visitor's question quickly. Another page with clearer definitions, direct answers, useful examples, strong authorship, and structured data may be easier for an answer engine to use.

This is good news for smaller businesses. You do not need the biggest marketing budget to produce the clearest, most credible answer.

At AEOvara, I approach AEO as a practical visibility discipline: remove ambiguity, show real expertise, and make important facts machine-readable without making the content sound robotic.

Start with real customer questions

The best AEO content usually starts in conversations, not keyword tools.

Listen to the questions prospects ask before buying:

How much does an accountant cost for a limited company in Finland?
When should I outsource payroll?
Which web agency is suitable for a small business in Lappeenranta?
What is included in a technical SEO audit?
How does Finnish VAT work for a new company?
Is service A better than service B for my situation?

These questions reveal intent. They also give you natural headings and concise answers that search engines and AI systems can understand.

Useful question patterns include how much, best, how to, near me, what is, compare, why choose, and when should I.

Answer the question immediately, then add context, evidence, exceptions, and a clear next step.

Build topic authority, not a pile of random posts

One strong page is helpful. A connected group of genuinely useful pages is much stronger.

Imagine an accounting firm. Its main service page could link to focused resources about:

VAT in Finland
Payroll outsourcing
Limited-company taxation
Accounting prices
Changing accountants

Together, these pages form a topic hub. Internal links clarify how the subjects relate, while consistent authorship and business information reinforce the entity behind the advice.

The goal is not to manufacture dozens of thin articles. It is to cover the decisions your customer actually needs to make.

Use structured data with purpose

Schema markup gives machines explicit clues about your organisation and content. It can clarify who you are, what you offer, where you operate, who wrote the page, and how different pages relate.

Useful schema types for Finnish SMEs often include:

LocalBusiness
Organization
Service
FAQPage
Product
Review
Article
Person
BreadcrumbList

Structured data is not a magic ranking button, and it should never describe information the visitor cannot see. Think of it as a clean data layer that supports the page's visible claims.

For example, an article about a specialist topic becomes more trustworthy when the page clearly identifies the author, links that author to a credible profile, shows a publication date, and accurately marks those facts in Article and Person schema.

Win the local answer

Local AEO matters because many searches carry immediate commercial intent:

accountant near me
photographer in Tampere
construction company in Espoo
SEO consultant in Finland

Create dedicated location pages only when you can make them genuinely useful. A strong local page includes specific services, local examples, customer stories, practical contact information, testimonials, and clear reasons the service fits people in that area.

Avoid duplicating the same text and swapping only the city name. That creates pages for algorithms rather than people—and neither audience is likely to be impressed.

Your Google Business Profile matters too. Keep categories, services, opening hours, photos, reviews, and answers accurate. Consistent business information across your site and trusted profiles helps systems connect the dots.

Publish content formats that answer buying questions

Blog posts have their place, but many buying decisions happen on other page types.

FAQ blocks

Put a short answer first. Follow it with nuance, examples, or conditions. Do not hide every important answer behind an accordion if the content becomes difficult to access.

Comparison pages

A fair comparison earns trust. Explain who each option is for, where the trade-offs are, and when your own service is not the best fit.

Pricing pages

You do not always need a fixed price. A starting price, typical range, package structure, or list of cost drivers is far more useful than “contact us for a quote.”

Step-by-step guides

Use numbered steps, descriptive headings, checklists, and concrete examples. Make each step understandable even when an answer engine extracts it from the wider article.

Think multilingual from the start

Many Finnish businesses serve customers in Finnish and English, while Swedish may also be commercially important.

Do not rely on literal translation alone. Search behaviour, vocabulary, examples, and customer expectations vary by language. Each version should sound natural, answer the local audience's questions, and use the correct language signals and internal links.

A useful English page can also improve discoverability among international founders, visitors, and companies looking for Finnish expertise.

Get the technical foundations right

Excellent content cannot do its job if a site is slow, confusing, or difficult to crawl.

Check that your website:

Uses HTTPS
Works properly on mobile
Loads quickly and remains visually stable
Has clear navigation and internal links
Avoids broken pages and redirect chains
Provides descriptive page titles and headings
Includes crawlable text rather than placing essential information only inside images
Uses canonical URLs consistently

Technical quality supports trust. It also reduces the risk that search and AI systems miss or misunderstand your best material.

Demonstrate experience and trust

E-E-A-T is not a badge you add to the footer. It is the cumulative impression created by experience, expertise, authority, and trustworthiness.

Show who wrote the content and why they are qualified. Use real names, useful author profiles, transparent company details, and references where a claim needs support. Keep old advice updated.

Reviews are more helpful when they mention the service, result, location, or experience rather than offering a vague “great company.” Case studies become stronger when they explain the challenge, strategy, timeline, measurable outcome, and business context.

Specificity is evidence.

A practical 30-day AEO plan

Week 1: Find the gaps

Audit your most important service pages. List the questions customers ask, check whether the pages answer them directly, and improve unclear headings.

Week 2: Fix the technical layer

Review mobile usability, performance, crawlability, canonical URLs, internal links, and relevant schema markup.

Week 3: Strengthen commercial content

Add useful pricing information, comparisons, FAQs, author details, reviews, and evidence. Turn vague claims into concrete facts.

Week 4: Expand and measure

Improve important local or language-specific pages. Track impressions, qualified visits, conversions, branded searches, referrals, and appearances in AI answers for a repeatable set of prompts.

Do not treat one ChatGPT query as a scientific benchmark. Models, personalisation, wording, and live retrieval can change the result. Use a consistent prompt set and look for trends over time.

Common AEO mistakes

Several shortcuts cause more trouble than they solve:

Publishing thin pages for every keyword variation
Duplicating city pages
Hiding all pricing information
Ignoring mobile performance
Producing only top-of-funnel blog posts
Adding unsupported superlatives such as “the best”
Using schema that does not match visible content
Writing for machines until humans stop enjoying the page

The fix is refreshingly unglamorous: be clear, accurate, specific, and helpful.

The takeaway

AEO is not tomorrow's tactic. It is today's competitive edge.

The Finnish SMEs that win in 2026 will not simply rank. They will be selected, cited, trusted, and chosen. That requires better answers, stronger trust signals, smarter structure, and content built around real customer needs.

Customers rarely want ten links. They want one clear answer.

Make sure it can be yours.

If you want to see how I apply these ideas in practice, explore the AEOvara expert profile and methodology.

Frequently asked questions

Is AEO replacing SEO completely?

No. AEO builds on SEO. Technical discoverability and rankings remain important, while answer-focused structure helps AI systems understand and cite the content.

Can a small Finnish business compete with larger brands?

Yes. Smaller businesses can win specific questions through stronger local relevance, clearer expertise, first-hand evidence, and better answers.

How quickly can AEO produce results?

Some improvements can appear within weeks, especially for local or question-based searches, but durable visibility requires consistent measurement and refinement.

Does every Finnish SME need multilingual content?

No. Add languages when they match real customers. Finnish and English are a sensible combination for many firms; Swedish matters in relevant markets and regions.

What is the best first step?

Audit your main service pages. Add direct answers to genuine customer questions, clearer pricing or cost drivers, trustworthy authorship, and evidence that supports your claims.

About the author

I'm Jarno S, founder of AEOvara in Finland. I work hands-on with technical SEO, Answer Engine Optimization, Generative Engine Optimization, structured data, and AI-search visibility. My focus is simple: help businesses become easier for people to trust and easier for search and answer engines to understand.

AI Search Visibility Is Finally Measurable: A Developer’s Field Guide for 2026

Thu, 10 Sep 2026 09:43:10 +0000

If you have ever tried to measure whether ChatGPT, Gemini, Copilot, or an AI-powered search result “knows” a brand, you have probably discovered the technical term for the experience: a mess.

Traditional search gave us rankings, impressions, clicks, and a fairly tidy funnel. AI answers broke that tidy funnel into pieces. A brand can be mentioned without a link, cited without a click, or used as background evidence while another source gets the visible citation.

The good news is that this is no longer completely invisible.

In 2026, Google and Bing are giving site owners more control and more reporting around generative search. That does not make AI visibility easy, but it does make it measurable enough to stop arguing from screenshots and start building a repeatable benchmark.

I’m Jarno S., founder of AEOvara, a Finnish consultancy focused on Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), and LLM visibility. I have been testing how content is discovered, interpreted, and recommended by AI systems since 2022. Here is the practical framework I use when I want evidence instead of vibes.

The short answer

AI search visibility is the measurable presence of a brand, person, product, or URL inside AI-generated answers. Track it with a fixed set of prompts, citation data from search platforms, referral analytics, and page-level content checks. Measure mentions and citations separately, because they are not the same thing.

If you only remember one formula from this article, use this:

AI visibility = mention share + citation share + answer accuracy + qualified actions

No single tool gives you the whole picture. Your job is to combine several imperfect signals into one useful trend.

Why the old SEO dashboard is not enough

Imagine someone asks an answer engine:

What is the best deployment platform for a small TypeScript team?

The answer may mention five platforms, cite three pages, and send zero clicks. One company might be named first but not cited. Another might supply the benchmark data. A third might get the only click because its title promises a useful comparison.

A normal rank tracker sees almost none of that.

This is why “we rank number three” is becoming an incomplete sentence. Number three where? In a blue-link result, an AI Overview, an AI Mode follow-up, a Copilot answer, or a model response generated without live web retrieval?

Classic SEO still matters. Google’s own guidance says that the same foundations remain relevant for its AI features: crawlability, useful content, internal links, page experience, and accurate structured data. There is no magic AI-only meta tag that turns average content into a trusted source.

What has changed is the output you are optimizing for. You are no longer chasing only a position. You are trying to become a source an answer system can confidently retrieve, understand, and cite.

The four signals worth tracking

Mention share answers: How often is my brand included when the model discusses this topic?

Build a stable prompt set around real user intentions. For example, a developer-tool company might test:

Recommend CI tools for a five-person Node.js team.
Compare hosted error-monitoring platforms for a Next.js app.
What should I use to document a public REST API?
Which tools are good alternatives to [known competitor]?

Run the same prompts on a schedule. Record which brands appear, their order, the wording around them, and whether the answer expresses confidence or caution.

Do not change half the prompt set every week. A benchmark that moves with your mood is not a benchmark.

Citation share answers: How often does the system visibly reference my domain?

This is stricter than mention share. A model may know your company name from many sources while citing a review, a documentation page, or a community thread instead of your website.

Bing Webmaster Tools now has an AI Performance report for citations across Microsoft Copilot, Bing’s AI-generated summaries, and selected partner experiences. It includes cited pages and the grounding queries associated with them. That is useful because it connects the answer back to both a URL and an intent.

For Google, check the generative AI reporting and controls available in Search Console for your property, then compare the trend with ordinary Search performance. The important part is not whether the numbers from different platforms match. They will not. The important part is whether each platform’s own trend improves over time.

3. Answer accuracy

Visibility is not automatically good news.

If an assistant calls your open-source library abandoned, invents a pricing tier, or confuses your product with another company, you have an entity problem. Track factual accuracy as a first-class metric:

{
  "prompt_id": "alternatives-07",
  "brand_mentioned": true,
  "brand_cited": false,
  "facts_correct": 3,
  "facts_checked": 5,
  "accuracy": 0.6,
  "notes": "Old pricing and wrong framework support"
}

This small bit of structure turns “the answer looked weird” into something your team can investigate.

4. Qualified actions

Eventually, visibility has to do something useful.

Track referrals from AI and answer platforms in analytics, but do not stop there. Look at what those visitors do: documentation views, demo requests, sign-ups, package installs, newsletter subscriptions, or whatever matters for your product.

AI referrals are often low-volume and high-intent. Ten visitors who arrive with a specific implementation problem can be more valuable than a thousand accidental impressions.

Build a prompt benchmark like a test suite

Developers already have the right mental model for this. Treat your AI visibility benchmark like a regression suite.

Start with 30 to 50 prompts. Group them by intent:

Discovery: “What tools solve X?”
Comparison: “X vs Y for this use case”
Validation: “Is X reliable or secure?”
Implementation: “How do I do X?”
Troubleshooting: “Why does X fail when Y?”

Then add three levels of specificity:

broad category prompts;
use-case prompts;
prompts that include constraints such as team size, language, budget, or region.

Run the suite from a clean context. Store the date, platform, model or mode when visible, prompt, answer, citations, and result classification. Do not compare a logged-in personalized result with an anonymous result and pretend they are equivalent.

A simple schema is enough:

CREATE TABLE ai_visibility_runs (
  run_date DATE,
  engine TEXT,
  prompt_id TEXT,
  brand_mentioned BOOLEAN,
  cited_urls JSON,
  mention_position INTEGER,
  accuracy_score DECIMAL(3,2),
  answer_hash TEXT
);

The answer hash helps identify repeated answers without storing endless duplicate text. Keep the original answer where your policies allow it, because qualitative changes often explain the numbers.

Make pages easy to quote, not merely easy to crawl

Technical access is the entry ticket. Citation-worthy content is the performance.

An answer engine needs passages that survive being lifted out of context. Give important pages:

one clear topic and intent;
a concise definition near the top;
descriptive headings that resemble real questions;
factual claims supported by primary sources;
named authors with relevant experience;
visible publication and update dates;
examples, limitations, and edge cases;
consistent facts across your website and public profiles.

That last point is easy to underestimate. If your homepage says the product supports ten integrations, the docs list eight, and an old directory lists six, the model has to choose among conflicting facts. Consistency is an optimization tactic.

Use structured data where it accurately describes visible content, but do not turn schema into fan fiction. Marking every author as an award-winning global expert does not manufacture expertise. It manufactures a debugging session for whoever inherits the site.

E-E-A-T is mostly a publishing discipline

Experience, expertise, authoritativeness, and trust are often discussed as if they were four secret ranking switches. A more useful way to think about E-E-A-T is as a publishing checklist.

Can a reader tell who wrote the page? Does the author have relevant first-hand experience? Are claims supported? Can someone verify the organization behind the advice? Is the content maintained? Are commercial interests obvious?

At AEOvara, my practical rule is simple: make the evidence visible. Do not merely say that you know a topic. Show the test, the decision, the result, and what did not work.

That principle helped shape the methodology I now use professionally. I first applied it while building and growing my own website: better semantic structure, direct answers, entity consistency, and deliberate internal linking. The work produced real search growth and became a testing ground for AEOvara’s approach. You can read more about my background and the case study on my AEO and AI-search expert page.

First-hand detail is also pleasantly hard to commoditize. Anyone can publish “ten GEO tips.” Only you can explain why your caching strategy failed on Friday, which query exposed the problem, and what the graph looked like after the fix.

What about llms.txt?

Treat llms.txt as an experiment, not a visibility button.

It may become a useful convention for pointing AI systems toward important resources. It does not replace crawlable pages, strong information architecture, sitemaps, robots directives, or good documentation. If you add one, keep it accurate and small. Then measure whether supported crawlers actually request it.

The same rule applies to every fashionable AEO tactic: if it cannot be tied to a plausible retrieval mechanism and a measurable outcome, it belongs in the experiment column, not the strategy column.

A 30-day implementation plan

Week 1: Establish the baseline

Choose your engines, define 30 to 50 prompts, and record mentions, citations, accuracy, and referral traffic. Export current data from Google Search Console, Bing Webmaster Tools, and analytics.

Week 2: Fix identity and access

Check crawlability, canonical URLs, robots rules, sitemaps, JavaScript rendering, author pages, organization details, and conflicting brand facts. Make sure your best content is reachable through normal internal links.

Week 3: Upgrade five source pages

Pick pages that already have impressions, backlinks, citations, or qualified traffic. Improve the direct answer, evidence, headings, examples, authorship, update date, and internal links. Five serious upgrades usually teach you more than fifty synthetic articles.

Week 4: Re-run and annotate

Run the same benchmark again. Annotate changes by URL and date. Look for direction, not miracles. AI answers are probabilistic, so one missing mention is noise; a repeated decline across related prompts is a signal.

Common mistakes I would avoid

Counting every brand mention as a win. A recommendation, a neutral mention, and a warning are different outcomes.

Publishing generic FAQ spam. Good answer formatting helps. Empty answers written only to occupy query space do not.

Changing content without logging the change. If visibility rises, you need to know what might have caused it.

Trusting one model run. Sample repeatedly and compare trends.

Ignoring the pages that models already cite. Those URLs are telling you what the system considers useful. Study their format and strengthen adjacent coverage.

Measuring only clicks. Zero-click visibility can influence later branded searches, direct visits, and purchase decisions.

Frequently asked questions

Is AEO different from SEO?

Yes, but they overlap heavily. SEO improves discovery and performance in search systems. AEO focuses on making content suitable for direct answers, citations, and recommendations. Strong technical SEO and people-first content remain the foundation.

Can you guarantee that an AI engine will cite a page?

No. Anyone who promises a guaranteed citation is selling certainty that the systems do not provide. You can improve eligibility, clarity, authority signals, and evidence; the final answer remains query- and system-dependent.

How often should an AI visibility benchmark run?

Monthly is a sensible starting point for most teams. Weekly runs can help during a migration or focused experiment, but daily testing often creates more noise than insight.

The useful conclusion

AI search optimization is becoming less mystical because the measurement layer is finally catching up.

You do not need to predict every model. Build a stable prompt benchmark. Separate mentions from citations. Score factual accuracy. Watch qualified actions. Improve the pages that already show signs of trust. Then repeat.

That is not a hack. It is an engineering loop — and it is much more reliable than hoping an AI chatbot says something nice about your brand.

About me: I’m Jarno S., founder of AEOvara. I specialize in AEO, GEO, AI SEO, and LLM visibility, and I’ve been testing AI-search content optimization hands-on since 2022.

How to Build a Repeatable AI Search Visibility Benchmark

Tue, 08 Sep 2026 13:42:54 +0000

How to Build a Repeatable AI Search Visibility Benchmark

By Jarno Saarimies, Founder of AEOvara · Last updated: September 8, 2026 · 9 min read

Your buyers are already asking ChatGPT, Perplexity, and Google AI Overviews who to trust. Google says AI Overviews now reach 2 billion monthly users, and industry tracking puts them on roughly 48% of all queries — up 58% year over year (BrightEdge, February 2026). Meanwhile, Seer Interactive's longitudinal study found organic CTR on AI Overview queries dropped from 1.76% to 0.61% between June 2024 and September 2025.

Here's the problem: your analytics stack is blind to all of it. If ChatGPT recommends a competitor and never mentions you, nothing gets logged. No click, no impression, no bounce rate. The opportunity just evaporates.

A benchmark fixes this. Not a one-off audit you run once and forget — a repeatable measurement system that tells you, month after month, whether you're becoming more or less visible inside AI answers.

This guide shows you exactly how to build one, starting with a free spreadsheet method and scaling to automated tracking when the data demands it.

What "AI search visibility" actually means (define it before you measure it)

Before you measure anything, agree on what you're measuring. Five metrics carry the weight:

Mention rate — the percentage of tested prompts where your brand appears at all. This is your headline number.
Share of Voice (SoV) — your mentions as a percentage of all tracked-brand mentions in the same answers. Peec AI's own documentation shows why this matters: a brand mentioned 4 times against a competitor's 12 mentions has high visibility but only 25% SoV. Visibility is presence. SoV is competitive prominence.
Citation rate — whether the answer actually links to you, not just names you. Track this separately: on ChatGPT most mentions carry no link, while Perplexity links almost everything.
Position and prominence — where you land in the answer. AI engines tend to treat the first-named brand as the default recommendation. Buried in the last paragraph is not the same as leading the list.
Sentiment and accuracy — not just whether you appear, but how you're described. A confident but outdated mention (wrong pricing, wrong positioning) can cost you the deal.

Step 1: Build a frozen prompt corpus

Pick 10–15 prompts that real buyers would actually ask. Not keywords — questions. Organize them into five categories:

Prompt type	What it reveals	Template
Category / discovery	Whether you're in the consideration set at all	"What are the best [category] for [use case]?"
Comparison	Positioning against a named rival	"How does [your brand] compare to [competitor]?"
Persona / use-case	Whether you're recommended in your buyer's context	"What should a [role] look for when choosing a [category]?"
Problem / jobs-to-be-done	Top-of-funnel presence before your category is named	"How do I [problem]?"
Brand-direct / reputation	Accuracy and sentiment	"What is [your brand]?" / "Is [your brand] any good?"

This list stays mostly frozen. This matters more than most people realize. If you change the prompts every month, you're not tracking visibility — you're running random tests. You can add prompts once a quarter, but the core set must stay stable or your trend data is meaningless.

Step 2: Choose your engines and cadence

Measure each platform separately, because a strong ChatGPT score tells you nothing about Perplexity or AI Overviews. At minimum: ChatGPT, Perplexity, Gemini, and Google Search with AI Overviews.

On cadence: weekly tracking is the standard for most teams (LLM Pulse and similar platforms default to this). Daily tracking makes sense for product launches or volatile categories — finance shows the highest citation volatility of any vertical. Monthly is the absolute floor for a manual process.

Run each prompt 3–5 times per engine. Generative answers are probabilistic — the same question names different brands on different runs. One query is an anecdote, not a measurement. Research on robust benchmarks suggests 500–2,000 prompts per week for statistically stable values (SUMAX Research, 2026); for a manual baseline, 10–15 prompts × multiple runs × 3–4 engines is a realistic starting point.

Step 3: Run the manual baseline (this afternoon, free)

You don't need a platform yet. You need a spreadsheet and about an hour.

Run your prompt set through each engine — same wording, same account settings where possible. For each response, log:

Query text
Platform and date
Mentioned? (Yes/No)
Position (1st named, 2nd, 3rd…)
Cited with a link? (Yes/No)
Competitors named
Sources cited in the answer
Accuracy/sentiment notes
Screenshot

Screenshot everything. AI results move around. Something that shows up this month may disappear next month. If you want to understand trends, you need receipts.

The last column is usually the most useful. If Perplexity or another engine shows sources, those sources tell you where the answer gets its confidence. Sometimes the fix isn't "write another blog post" — it's getting your company described better on the third-party pages the AI is already reading.

Step 4: Calculate your benchmark numbers

Four formulas turn raw logs into a benchmark:

Mention rate = prompts where you appear ÷ total prompts run × 100
Citation rate = prompts with a link to you ÷ total prompts × 100
Share of Voice = your mentions ÷ all tracked-brand mentions × 100
Position score = average position when mentioned (lower is better)

Example: mentioned in 5 of 12 prompts on Perplexity, with 4 of those including a link. Mention rate: 42%. Citation rate: 33%. Your competitor mentioned in 9 of 12 with 7 links. SoV: 5 ÷ 14 = 36%.

That gives you a real baseline — and usually a few uncomfortable surprises. One thing to internalize early: position in a single answer is mostly noise; frequency across many runs is the signal (as SparkToro's analysis showed). Track how often you appear, not where you landed once.

Step 5: Add context metrics the spreadsheet can't see

A prompt log tells you that you're visible or invisible. Three additional signals tell you whether it matters:

AI referral traffic. Expect attribution to undercount. Add a "How did you hear about us?" option with an AI assistant choice to your intake forms, because most AI mentions never produce a trackable click.

Branded search lift. Growth in people searching your company name often correlates with AI exposure, per Percepture's GEO research.

Conversion quality. The traffic you do get is disproportionately valuable: Ahrefs' own data showed AI-referred visitors converting at 23x the rate of traditional organic visitors (0.5% of traffic generated 12.1% of signups, June 2025). Semrush found a similar 4.4x pattern. Small numbers, high intent.

Step 6: Automate when manual stops scaling

Twenty-five prompts run five times each is 125 data points — enough for a baseline, not enough for stable rates. That's when a platform earns its money. Roughly priced landscape as of mid-2026:

Otterly AI — from $29/month (15 prompts). The low-cost serious entry point; weekly refresh.
Peec AI — from $95/month (50 prompts). Clean visibility, position, sentiment, and SoV dashboards; strong for structured reporting.
Semrush AI Visibility Toolkit — from $99/month standalone. Low friction if you're already on Semrush.
Ahrefs Brand Radar — from $199/month per AI-platform index (requires an Ahrefs base plan from $129/month). Prompt methodology anchored to search data.
Profound — enterprise positioning, deep analytics, most advanced features behind custom pricing.

Whichever tool you pick, demand these capabilities: multi-platform tracking (ChatGPT, Perplexity, Gemini, AI Overviews, Claude), citation/source URL analysis — not just mention counts — competitor benchmarking, historical trends, and transparency about prompt volume methodology.

The seven mistakes that break benchmarks

Chasing rank position instead of visibility rate. Position swings wildly run to run; frequency is the stable metric.
Measuring only one engine. You're optimizing blind.
Counting mentions but ignoring citations (or vice versa). They're different metrics doing different jobs.
Ignoring sentiment and accuracy. Being named with wrong pricing isn't a win.
Running each prompt once. Anecdote, not measurement.
Expecting clean click attribution. Analytics will always undercount AI's influence.
Testing once and moving on. Visibility drifts as models update and content ages. Only a fixed cadence reveals what's actually moving.

The bottom line

You can't manage what you can't see, and traditional analytics can't see AI answers. The process is unglamorous: freeze a prompt set, run it repeatedly across engines, log mentions and citations separately, review monthly, automate when the volume outgrows your spreadsheet. Do that consistently and "how visible are we in AI?" stops being a worry and becomes a number you can move.

FAQ

What is an AI search visibility benchmark?
An AI search visibility benchmark is a fixed set of buyer-relevant prompts that you run repeatedly across AI search engines (ChatGPT, Perplexity, Gemini, Google AI Overviews) to measure how often and how accurately your brand appears in AI-generated answers over time. It turns "are we visible in AI?" from a vague worry into a trackable number with a trend line.

How many prompts do I need for a reliable benchmark?
For a manual baseline, 10–15 prompts across five categories (category, comparison, persona, problem, brand-direct), run 3–5 times per engine, gives directional data. For statistically stable visibility rates, research suggests 500–2,000 prompts per week — which is where automated platforms become necessary.

What's the difference between mentions and citations?
A mention means the AI names your brand in the answer text. A citation means it links to your website as a source. Mentions build brand association; citations drive trackable traffic. On ChatGPT most mentions carry no link, while Perplexity links almost everything, so the two metrics must be tracked separately.

How often should I re-run my benchmark?
Weekly is the standard cadence for automated tracking; monthly is the minimum for manual runs. Re-run the exact same prompt set each time — changing prompts between runs destroys comparability. Also re-baseline after major model updates, since visibility can shift overnight when models change.

Can I build an AI visibility benchmark for free?
Yes. A spreadsheet, 10–15 frozen prompts, and one hour a month covers the fundamentals — mention rate, citation rate, position, competitors, and cited sources. Paid tools earn their cost when you need scale, daily cadence, sentiment scoring, and source-level analysis across hundreds of prompts.

Do I still need traditional SEO tracking?
Yes. Ranking on page one remains correlated with AI citations — Ahrefs found 38% of AI Overview-cited pages also rank in the top 10 — but the relationship is weakening fast (down from ~76% in mid-2024). The two systems measure different things: blue-link position versus presence inside the answer. You need both.

Jarno Saarimies is the founder of AEOvara, a Finnish consultancy specializing in AI search optimization (AEO/GEO/LLMO) — helping brands get found, cited, and recommended inside AI-generated answers. A professional photographer since 2013, Jarno runs Kuvaajankulma, a photography studio in Lappeenranta, Finland, and applies the same evidence-first mindset to search: no hype, no guesswork — measure what actually matters. AEOvara Jarno S
AEOvara BLOG

Is SEO Dead? What AI Search Really Means for Website Traffic

Wed, 19 Aug 2026 10:03:31 +0000

Is SEO Dead? What AI Search Really Means for Website Traffic

A few months ago, I noticed something odd in my traffic data. A page that used to bring in steady clicks for a simple "what is X" question had quietly dropped. Nothing had changed on the page. No penalty, no broken link, no algorithm update I could point to.

When I searched the same query myself, I found the answer. Google was answering it directly at the top of the page, before any website even showed up.

That's when the question everyone in SEO is quietly asking hit me too: is SEO dead?

I spent time digging into this. Testing queries, checking AI Overviews, running the same searches through ChatGPT, Gemini, and Perplexity, and comparing what happened to traffic on pages that used to rank well. Here's what I actually found.

Try this yourself before you read on

Pick one question you rank for. Search it on Google right now.

Does an AI Overview answer it completely, with no need to click anywhere?
Or does the answer only get you started, and you still want to compare, verify, or go deeper?

Keep your answer in mind. It matters for what comes next.

The short answer: No. But it's not the same game anymore.

Google still handles roughly 90% of all search traffic. That hasn't changed. What's changed is what kind of content gets clicked versus what gets swallowed by an AI-generated answer at the top of the page.

If your content exists to answer a simple factual question, like "what year was X founded" or "how many ounces in a cup," AI Overviews are now doing that job before anyone reaches your site. That traffic isn't coming back. No amount of "better SEO" fixes that, because the problem isn't your ranking. It's that there's no click to earn anymore.

But that's only one slice of search. The moment a query gets more specific, more personal, or higher intent, like "best CRM for a 5-person team" or "compare these two tools for my use case," people still click through. They still want to read a real answer, compare options, and see evidence. AI Overviews are good at summarizing consensus. They're not good at replacing judgment.

What's actually dying

It's not SEO. It's a specific kind of content:

Thin articles written only to rank for a keyword, with no real insight
Content that just repeats what's already on page one
Listicles with no original testing, opinion, or experience behind them
Anything that reads like it was generated to satisfy an algorithm, not a person

That content was already losing ground before AI Overviews existed. This just sped it up. Google's own guidance has pushed toward E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) for years. AI search just makes the gap between generic content and content with real experience behind it impossible to ignore.

What's rising in its place

Two disciplines have shown up alongside traditional SEO. Worth knowing even if you don't specialize in them yet.

AEO, Answer Engine Optimization. Structuring content so AI systems can pull a clear, direct answer from it. Clean headings, direct answers near the top, FAQ-style sections, and schema markup (FAQPage, HowTo, Article) that tells machines exactly what your content is about.

GEO, Generative Engine Optimization. Making sure your brand actually gets mentioned inside AI-generated answers, not just ranked in blue links. A different kind of visibility. You're not optimizing for a click anymore, you're optimizing for a citation.

Both sit on top of the same foundation SEO has always needed: fast, crawlable, well-structured, technically sound websites. Nothing about that has gone away. If anything, it matters more, because AI systems are even less forgiving of a broken or confusing site than a human skimming search results.

Quick self-check: are you ready for AI search?

Ask yourself these four questions honestly.

Does your most important page answer the core question in the first two sentences?
Do you have schema markup (FAQPage, HowTo, Article) on your key pages?
Does your content include something only you could write (real testing, real numbers, real experience)?
Can a crawler actually read your page, or does it rely on JavaScript to load the content?

If you answered "no" to two or more, that's exactly where to start.

What still works: no hype, just fundamentals

Testing this myself confirmed something I already believed. The fundamentals haven't moved.

Technical SEO. Core Web Vitals, crawlability, and a site that actually loads and renders properly. AI crawlers struggle with JavaScript-heavy pages the same way traditional crawlers do. If your content doesn't render, it doesn't get read, summarized, or cited.
Structured data. Schema markup isn't just an SEO checkbox anymore. It's how you tell both Google and AI systems exactly what your content is, in a format they can parse instantly.
Real experience content. Original insight, real testing, first-hand results. This is the one thing AI genuinely cannot replicate, and it's becoming the clearest differentiator between content that ranks and content that gets ignored.
Depth over volume. One well-researched, honestly written article now consistently outperforms five shallow ones. Quality has always mattered. It just used to be possible to get away without it. That's no longer true.

What's genuinely new

The one real shift is where visibility happens. Ranking #1 on Google used to be the finish line. Now there's a second finish line: does ChatGPT mention your brand when someone asks a relevant question? Does Perplexity cite your site as a source?

That's a new kind of visibility, and right now, almost nobody is tracking it. Which means there's real opportunity for anyone who starts paying attention early.

So, is SEO dead?

No. It's splitting into two related disciplines that increasingly overlap: getting found in traditional search, and getting cited inside AI-generated answers.

The sites that win in 2026 won't be the ones that "did more SEO." They'll be the ones that built genuinely useful, technically solid, honestly written content. The kind that was always going to hold up, no matter how search itself keeps changing.

The algorithm changed. The bar for good content didn't. If anything, it just got higher.

One question to leave you with: when's the last time you searched your own site's main keyword and actually looked at what showed up above your listing?

Website: devndespro.com | seo.devndespro.com

Your SaaS competitors aren't who you think they are.

Tue, 04 Aug 2026 11:30:14 +0000

The future belongs to original thinkers, not content factories.

In a post-2026 environment, the competitive set for most SaaS, mobile and web products has expanded far beyond the usual list of direct rivals.
Vibe-coding (rapid, high volume generation of functional software via AI assistance) has lowered the cost and time of shipping something that looks and mostly works enough to enter the market. The result is a flood of near substitutes.

The average product no longer competes mainly against a handful of
well funded or well designed incumbents; it competes against a
continuous stream of “good enough” alternatives that can be rigged up
in days or weeks.

AI systems themselves have become a major allocation layer for attention and trust. When models surface, rank, recommend or cite sources they tend to favor signals of authority, consistency and apparent longevity. Pure traffic volume or short term ranking games matter less than they did in classic web search. A digital papertrail of substantive, non spammy, non generic content that people can actually use or build on carries more weight than volume alone. Models are communicating with each other and their extreme sensitivity to commonality will quickly spread, causing your SaaS or content to never even surface in the answer reconstruction:

-The same recycled patterns the same phrasing the same surface level claims, show up as very low signals.
-Distinctive, verifiable or applied knowledge that is not sugarcoated to preserve conversions, stands out more.
In context, spam remains low value. The models treat it as such.

Commonality is corrosive when it is just stylistic or structural clones but shared facts and actual reproducible methods are not the problem they are the substrate, the base layer.
The distinction is between undifferentiated output and content that adds a usable increment of understanding or capability.

The practical implication is that differentiation now has to survive
both human and AI model evaluation, however, it's no secret
that in over 20+ years, humans have been #1 consumers of the Internet.
Today is August 4th, 2026, 04:00, AI agents are now the #1 consumers
of the world wide web.

Shipping faster is table stakes because many others can also ship fast. What compounds is a visible record of work that is specific, testable and hard to fully replicate or clone by pure generation.

Traffic still matters but it is being mediated by systems that discount pure volume and pure commonality. The products and creators that keep getting surfaced and retained are those that look less like interchangeable vibe coded app and tool instances and more like nodes with an actual trail of useful signal. Articles or written content are also passing the same filters, welcoming back the originality in human-first content, similar to what you've just read.

SEO, AEO, GEO & SERP Explained: The 2026 Guide to Search Visibility

Fri, 17 Jul 2026 04:25:35 +0000

SEO, AEO, GEO & SERP Explained: The 2026 Guide to Search Visibility

Search Visibility · Updated July 2026 · 10 min read · by Jacob Campbell

Search used to mean one thing: rank on Google. In 2026 it's four overlapping disciplines — SEO, the SERP, AEO and GEO — and together they decide whether real people and AI assistants ever find your site. The reassuring part is that they aren't competing strategies; they're layers built on the same foundation of fast, well-structured, trustworthy pages. This guide explains each term in plain English, shows how they connect, and why getting them right is one of the clearest signals of a healthy website.

Key facts

4 — Layers: SEO, SERP, AEO, GEO
~27% — of clicks go to the #1 SERP result
AI — Answer & generative engines now cite sources
1 base — All four rest on fast, structured pages

The four terms, in one minute

SEO (Search Engine Optimization) is the work of making your pages rank in search engines. The SERP (Search Engine Results Page) is the results page itself — the battlefield where rankings show up, alongside ads, snippets and AI answers. AEO (Answer Engine Optimization) is optimizing to be the direct answer a machine shows in a box or reads aloud. GEO (Generative Engine Optimization) is earning a mention inside AI-generated answers from tools like ChatGPT, Gemini and Google's AI Overviews. Four names, one goal: be found.

SEO — the foundation everything sits on

SEO is still the base layer. It covers technical health (fast load, clean code, crawlability), on-page relevance (titles, headings, content that matches intent) and authority (links and trust). Nothing else works without it: an answer engine can't quote a page it can't crawl, and a generative engine won't cite a source it doesn't trust. Get SEO right and you've built the foundation the other three layers stand on. See our guide to why sites don't rank for the common failure points.

The SERP — where visibility is won or lost

The SERP is no longer ten blue links. A single results page now blends paid ads, a local map pack, featured snippets, People Also Ask boxes, images, video and — increasingly — an AI Overview at the very top. Every one of those is a slot you can win or lose. Understanding the SERP means designing your pages for the specific features your keywords trigger, not just for an abstract 'position one.'

AEO — becoming the answer, not just a link

Answer Engine Optimization targets the boxes and voice results that hand users a direct answer: featured snippets, People Also Ask, and voice assistants. You win these by answering real questions concisely, structuring content with clear headings, and marking it up with schema (FAQ, HowTo, Article). When your page is the cleanest, best-structured answer to a question, the engine reads it back to the user — often with your name attached. Our schema markup guide covers the structured data that makes this possible.

GEO — getting cited inside AI answers

Generative Engine Optimization is the newest layer: earning a mention when ChatGPT, Gemini, Perplexity or Google's AI Overviews generate an answer. These systems favour sources that are clearly written, factually specific, well-structured and trusted across the web. In practice GEO rewards the same things good SEO and AEO do — clarity, structure, citations and authority — which is why a healthy site tends to appear in AI answers without gaming anything.

Why all four are a website-health signal

Here's the tie that binds them: every layer is powered by the same fundamentals — fast Core Web Vitals, clean crawlable HTML, structured data, clear content and genuine authority. A site that performs across SEO, the SERP, AEO and GEO is almost by definition technically healthy; a site that fails them is usually slow, bloated or thin. That's why search visibility is a health check, not a marketing add-on. Start with the technical base — passing Core Web Vitals — and the rest compounds.

Sources & further reading

Related services

Frequently asked questions

What's the difference between SEO, AEO and GEO?

SEO optimizes to rank in search results, AEO optimizes to be the direct answer in snippets and voice results, and GEO optimizes to be cited inside AI-generated answers. All three build on the same foundation of fast, well-structured, trustworthy pages.

What is a SERP?

The SERP is the Search Engine Results Page — the page you see after a search. In 2026 it mixes ads, a map pack, featured snippets, People Also Ask, images, video and an AI Overview, not just ten links.

Is SEO still relevant with AI search?

Yes — more than ever. AI answer and generative engines can only quote and cite pages they can crawl, understand and trust, which is exactly what SEO delivers. SEO is the foundation AEO and GEO stand on.

How do I optimize for AI Overviews and ChatGPT?

Write clear, factually specific content, structure it with proper headings and schema, and earn genuine authority. Generative engines favour well-structured, trustworthy sources — the same qualities that win featured snippets.

How does search visibility relate to website health?

Every search layer depends on speed, clean code, structured data and clarity. A site that ranks, wins snippets and gets cited by AI is almost always technically healthy; poor visibility usually signals a slow or bloated site underneath.

Want a site built to win every search layer?

Every BuiltToWinWeb site ships fast, structured and schema-ready — the foundation for SEO, AEO and GEO — for one flat fee.

Get my free quote

Programmatic AI Discovery: Automating the LLMs.txt Standard with a Sitemap-to-LLM Loop

Sat, 20 Jun 2026 15:08:26 +0000

Hey developers and tech founders,

We’ve all seen the rapid adoption of the /llms.txt standard. It is an elegant, open-source way to declare our system documentation, core application routes, and API endpoints for clean LLM crawler ingestion (GPTBot, ClaudeBot, PerplexityBot, etc.).

But as developers, we hate manual maintenance. If you are building a rapidly growing product, manually updating links inside a static plain-text Markdown file every time you deploy a new API route, document, or blog post is a terrible anti-pattern.

Here is how you can bypass the operational overhead entirely by implementing an automated Sitemap-to-LLM Loop.

The Core Concept: The XML-to-AI Bridge

Many developers overlook the fact that modern LLMs are already highly proficient at parsing XML sitemaps natively. They do not need you to translate every single deep URL into a manual Markdown string. They can read and extract paths directly from your standard, dynamically updated /sitemap.xml.

By structuring your /llms.txt to handle the high-level semantic introduction (who you are, what your product does) and pointing AI crawlers directly to your dynamic XML sitemap, you create a zero-maintenance loop. Your backend automatically updates the sitemap on deploy, and all major AI bots crawl your changes automatically.

The Code: How to Serve It Dynamically

If you are running a custom Node.js/Express backend, you don't need to host a static file. You can serve your /llms.txt dynamically as a flat text file with the correct MIME type:

const express = require('express');
const app = express();

const LLMS_TEXT_CONTENT = `# SEOSiri Technical Documentation & Engineering Ecosystem

SEOSiri is a B2B digital engineering consultancy and technical SEO platform 1.

Core Services & Blueprints

Sovereign CapEx Engine: Web-based capital expenditure modeling, NPV/IRR forecasting, and asset amortization tracking [1.1.1].
Biometric IoT Bridge: Secure Flutter biometric-to-hardware TLS/MQTT edge gateway framework.
Precision Agriculture E-Noses: Machine-learning-driven agricultural sensor telemetry and environmental monitoring pipeline.

Dynamic Site Map & Full Index

Dynamic XML Sitemap: Our real-time updated, machine-readable sitemap containing all publication and category URLs. AI crawlers can use this sitemap to discover and index all deeper historical technical articles 1.`;

// Serve llms.txt with the correct text/plain MIME type
app.get('/llms.txt', (req, res) => {
res.set('Content-Type', 'text/plain');
res.send(LLMS_TEXT_CONTENT);
});

The Proven Results
We deployed this exact, low-overhead system on our platform, SEOSiri. Verified by our custom AI visibility dashboards, this automated architecture has generated 258 total AI citations and a 21.24% Share of Authority (SoA) across Microsoft Copilot and partners in just 15 days.

If you want to implement this automated loop across different environments, we have published the complete technical roadmap. Our guide covers step-by-step instructions for WordPress, Shopify, Ghost, Webflow, Wix, Squarespace, Blogger, Node.js, Django, and custom HTML/CSS sites:

👉 Read the Full Automated LLMs.txt Integration Guide on SEOSiri

I'd love to hear from other backend developers and full-stack builders here:

How are you structuring your documentation for AI-first search? Have you integrated a programmatic /llms.txt route into your deployment pipelines, or are you still relying on manual updates? Let's discuss!

Magento AI Brand Visibility: Does ChatGPT Recommend Your Store?

Sat, 13 Jun 2026 07:26:48 +0000

Originally published at angeo.dev. This is a cross-post; the canonical version lives there.

Short answer: when your customers ask ChatGPT, Claude or Perplexity "where should I buy this?", you currently have no idea whether your store appears in the reply. Angeo AEO Brand Visibility is a free, open-source Magento 2 module that measures exactly that: it runs brand-probing prompts across the five major AI models and scores your real-world AI brand visibility from 0 to 100.

Search no longer means only Google. More buying research now starts inside an AI assistant that returns one synthesized answer naming a few stores. AI Engine Optimization (AEO) is the work of making sure your brand is one of those names — and this module tells you, in minutes, whether it is.

Why AI brand visibility matters for Magento merchants

AI answers are winner-takes-most. Where a search results page lists ten links, an AI assistant usually names two or three stores. If you're not in that shortlist, you're invisible to that buyer — there is no "page two". Traditional SEO tools can't see this: they measure rankings, backlinks and crawl health, not what a model actually says when prompted. That blind spot is what AEO Brand Visibility closes.

How Magento AI brand visibility tracking works

You set your brand name, domain, category and a few top products. The module sends natural shopping prompts — "What are the best online stores to buy {category}?", "Tell me about {brand}", "Compare {brand} with similar stores" — to every enabled AI provider. Each response is analysed for five signals, scored, cached and saved to a history log so you can watch the trend over time.

Supported AI providers

Provider	Why it's included	Free tier?
ChatGPT (OpenAI)	The most-used assistant; the default benchmark for AI recall.	No
Claude (Anthropic)	Strong reasoning; common for research-style queries.	No
Perplexity	Live web search — best signal for what the internet says about you now.	No
Gemini (Google)	Tied into Google's ecosystem; generous free tier.	Yes
Groq	Fast, free Llama hosting — ideal for zero-cost testing.	Yes

Start with the free providers (Gemini and Groq) to validate your setup at zero cost, then add paid providers for the most commercially relevant picture.

The five visibility signals

Mentioned — your brand or a configured alias appears anywhere in the answer.
Recommended — the model actively suggests you, or you appear in a ranked list.
URL cited — your domain is referenced in the response, the strongest trust signal.
First result — you appear in the first quarter of the answer, i.e. top-of-mind.
Positive sentiment — favourable language sits close to your mention.

Each signal is weighted and combined into a single 0–100 score with an A–F grade.

What you get in the Magento admin

Run Audit — one click queries every enabled model and shows live scores, per-provider chips and signal rates.
Statistics & trend chart — average, best and worst scores plus a score-over-time graph from your fresh (non-cached) runs.
Audit History — your recent runs, each expandable to the full prompt-by-prompt detail.
Action Plan — a prioritised, time-bounded roadmap of what to fix first.
Single Query Tester — send one prompt to one provider and inspect the raw response.

Automate it: CLI, cron and CI gates

AI brand visibility drifts as content, reviews and the web change, so the module ships a CLI command and a cron job — and it can fail a CI build if your score drops below a threshold.

# Run a full audit across all enabled providers
bin/magento angeo:aeo:brand-visibility

# Force fresh queries (bypass the cache)
bin/magento angeo:aeo:brand-visibility --refresh

# Test a single provider / prompt
bin/magento angeo:aeo:brand-visibility --provider=perplexity --prompt=brand_direct

# Machine-readable output for dashboards
bin/magento angeo:aeo:brand-visibility --format=json

# Gate a pipeline: exit 1 if the score falls below 60
bin/magento angeo:aeo:brand-visibility --fail-on=60

Built to extend angeo/module-aeo-audit

AEO Brand Visibility plugs into angeo/module-aeo-audit as a live-signal checker, sitting alongside the 15 built-in technical checks (robots.txt, llms.txt, structured data and more). Your brand-recall score becomes part of the same unified AEO report — one command, one dashboard, technical and real-world signals together.

Security & privacy

Provider API keys are stored with Magento's encrypted backend model and never written to logs. Outbound calls are HTTPS-only and don't follow redirects, admin endpoints are protected by ACL and form keys, and all AI-provider text is escaped before rendering. Serialization uses Magento's SerializerInterface throughout.

Installation

composer require angeo/module-aeo-brand-visibility
bin/magento setup:upgrade
bin/magento setup:di:compile
bin/magento cache:flush

Then open Stores → Configuration → Angeo AEO → Brand Visibility, add at least one provider API key, set your brand name and domain, and hit Run Audit.

How to improve your AI brand visibility score

Low mention rate? Publish an llms.txt file (angeo/module-llms-txt) and keep your store name consistent across every page.
Domain not cited? Strengthen backlinks and reference your canonical URL in structured data and llms.txt.
Not being recommended? Improve product content quality and implement review schema.
Appearing late in answers? Build topical authority with category-focused content.

FAQ

Is the module free? Yes — MIT licensed. You only pay for the AI provider usage you choose, and you can run it entirely free using the Gemini and Groq free tiers.

Which AI models does it check? ChatGPT, Claude, Perplexity, Gemini and Groq. Enable any combination.

Will audits cost a lot in API fees? Not if configured sensibly. Results are cached, you control how many prompts run, and total queries equal enabled providers × active prompts.

What Magento versions are supported? 2.4.6, 2.4.7 and 2.4.8 (Adobe Commerce and Mage-OS), PHP 8.2–8.4. Requires angeo/module-aeo-audit v3.0+.

How is this different from SEO tools? SEO tools measure rankings and crawlability. This measures what AI assistants actually say when asked shopping questions.

AI assistants are already recommending stores in your category — the only question is whether they're recommending you. Install angeo/module-aeo-brand-visibility, run your first audit, and find out in minutes. You can also run a free 30-second AEO audit.

Originally published at angeo.dev.

GEO & AEO for Technical Founders: Getting Your Code, SaaS, and Web Apps Cited by AI Search

Wed, 10 Jun 2026 15:18:45 +0000

For software developers, SaaS builders, and technical architects, building a great product is only half the battle. If your web application, developer utility, custom extension, or open-source package isn't properly indexed and understood by AI search models, you are missing out on high-intent B2B adoption.

When enterprise buyers ask ChatGPT, Gemini, or Perplexity for the most secure and scalable web solutions, your systems need to surface as the trusted source.

We have published a deep dive on how we are leveraging Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) to help developers, startups, and technical brands secure authoritative citations inside AI-first search ecosystems.

Read the full breakdown and learn how to showcase your software platforms, web apps, or technical systems on our platform:

https://www.seosiri.com/2026/06/tech-interview-digital-pr-feature-seosiri.html

Everyone's telling you to add llms.txt for AI search. The 2026 data says it gets 0.1% of crawler tra

Sat, 06 Jun 2026 19:10:35 +0000

I'll save you the suspense: yes, you should still ship llms.txt on your Magento store. But not for the reason every "AEO guide" is selling you — and if you're shipping it instead of fixing the things that actually get you cited in ChatGPT, you've optimized the wrong file.

This is the contrarian version of the AEO conversation, with the numbers attached. I've been measuring this across real Magento 2 stores, and the gap between the marketing and the server logs is wide.

The claim being sold

The pitch goes: AI is the new search, llms.txt is the new robots.txt, add one and ChatGPT/Claude/Perplexity will start recommending you. Vendors are charging for modules that generate it. Plugins are auto-shipping it.

What the logs actually show

The independent 2026 data is blunt:

A 300,000-domain study found roughly a 10% adoption rate for llms.txt. Among the 50 most AI-cited domains, exactly one had the file.
One vendor instrumented a domain for 90 days: 62,100 AI bot visits, 84 of them hit llms.txt. That's 0.1%. They removed it from their GEO audit checklist afterward.
Another monitoring 500M+ AI bot events found the same pattern — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, and Google-Extended overwhelmingly skip the file and crawl your HTML directly.
OpenAI, Anthropic, and Perplexity have not committed to reading it automatically. Google's John Mueller has said major crawlers don't prioritize it over standard HTML.

If your goal is "get cited more in AI answers," llms.txt is not the lever. Full stop.

So why ship it anyway?

Because it's a Business-to-Agent (B2A) play, not an SEO play — and on that framing it's worth the half-day:

Agentic and IDE-style tooling already fetches it. The agent ecosystem is where this pattern is mature, and agentic commerce is heading the same way.
Cheap insurance. Conventions like this routinely get published before platforms formally commit — robots.txt itself predated official search-engine support. Being already-correct when adoption crosses the threshold costs almost nothing now.
The exercise is clarifying. Writing one authoritative paragraph about your store forces the question "what would I want a model to say about us?" — which most catalogs have never answered.

Ship it. Then stop touching it. A stale llms.txt is worse than none — wrong stock or dead policy links train models to distrust your feed.

Where the citations actually come from

Here's the part the llms.txt hype distracts from. On Magento specifically, two things move the needle far more, and both ship off by default:

1. robots.txt — you're probably blocking the AI fleet by accident. A robots.txt written for Google in 2019 says nothing about ~10 distinct AI bots, and hardening configs that block "unknown" bots now block the crawlers feeding ChatGPT and Claude. Note: Google-Extended is not Googlebot — many sites blocked it in the 2023–24 anti-scraper panic and killed their Google-AI visibility without touching their ranking.

User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: Google-Extended
Allow: /

2. JSON-LD with a live offers.availability. Default Magento 2 emits microdata, not JSON-LD — the format AI extraction is worst at. And the killer omission: offers.availability is frequently missing, and ChatGPT Shopping skips a product with no availability signal entirely. It's assembling a purchasable recommendation; an item it can't confirm is in stock is a liability, so it drops you silently.

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "Cast Iron Skillet 12\"",
  "sku": "CAST-SKILLET-12",
  "offers": {
    "@type": "Offer",
    "price": "59.00",
    "priceCurrency": "EUR",
    "availability": "https://schema.org/InStock"
  }
}
</script>

Two Magento traps that waste people's time:

Schema injected via GTM is invisible — AI crawlers generally don't run JavaScript, so it validates in DevTools and is unreadable to the bot. Server-render it.
Hyvä ships zero product schema by default. Migrated for performance? You may have silently dropped your structured data.

Confirm it's in the raw HTML, not just the rendered DOM:

curl -s https://yourstore.com/your-product | grep -A20 'application/ld+json'

Empty output = the AI layer sees nothing, no matter how green your audit looks.

And the part that's actually about revenue: ACP vs UCP

While everyone argues about llms.txt, the real 2026 shift is AI moving from discovery to transaction. Two protocols:

ACP (OpenAI + Stripe) — live in ChatGPT since late 2025, now a discovery-and-feed role with delegated payment tokens and a transaction fee on Instant Checkout orders.
UCP (Google + Shopify + 20-plus partners) — NRF Jan 2026, expanded May 2026; protocol-agnostic, spans discovery → checkout → post-purchase across Google AI Mode, Gemini, YouTube, Gmail.

The take that matters: it's not "pick one." They hit different agent ecosystems and dual-protocol merchants reportedly see meaningfully more agentic traffic. Most groundwork — clean schema, accurate live feed, defined policies — is shared and survives spec churn. Shopify gets this by partnership default; Magento merchants own the stack and configure it themselves.

The actual priority order

Fix robots.txt. (~20 min, highest leverage)
Server-rendered JSON-LD with live offers.availability.
curl verify — not in raw HTML = invisible.
Then ship llms.txt as B2A insurance, and leave it.
Build the shared feed/policy substrate, attach ACP then UCP.

The headline stands: llms.txt is the most over-marketed item in AEO right now. Ship it cheaply, ignore the hype, and spend your real hours on the bots-can-crawl-you and schema-is-server-rendered problems. That's where the citations live.

Measured across Magento 2 stores at angeo.dev; open-source AEO audit CLI + llms.txt generator on Packagist. The adoption/crawler figures are from independent 2026 studies — verify current spec versions as you go, this moves monthly.

Have you pulled your own AI-bot logs on a Magento store? I'd love to see whether your llms.txt hit-rate matches the 0.1% — post your numbers below.

Your Website Ranks #1 on Google. ChatGPT Has Never Heard of You

Fri, 29 May 2026 13:25:17 +0000

title: Your Website Ranks #1 on Google. ChatGPT Has Never Heard of You.
published: false
tags: seo, ai, webdev, productivity
cover_image: https://devndespro.com/og-aeo.png

canonical_url: https://dev.to/karthic2914/your-website-ranks-1-on-google-chatgpt-has-never-heard-of-you

Google ranked your site #1. Traffic is good. Life is fine.
Then someone opens ChatGPT and asks for a recommendation in your niche.
Your name does not appear. A competitor you have never heard of gets cited.
The user never even opens a browser.

This is the gap between SEO and AEO - and it is growing every single day.

What is AEO?

AEO stands for Answer Engine Optimization.

It is the practice of optimizing your website not just for traditional search engines, but for AI-powered answer engines:

Platform	What it does
ChatGPT	Uses Bing index + web browsing to answer questions
Perplexity	Heavily cites Reddit, forums, and authoritative sources
Google AI Overviews	Extracts direct answers from structured on-page content
Gemini	Uses Google's knowledge graph + E-E-A-T signals

SEO asks: Can Google find and rank my page?

AEO asks: Can an AI find, understand, trust, and cite my page?

These are very different questions.

How AI Engines Decide What to Cite

When you ask ChatGPT or Perplexity a question, they do not just search Google. They evaluate multiple signals:

✅ Bing indexing         - ChatGPT uses Bing. Not indexed = invisible to ChatGPT.
✅ Reddit presence       - Perplexity weights Reddit heavily as a citation source.
✅ Structured content    - FAQ schema, HowTo schema, concise answer paragraphs.
✅ Author entity         - Named author with credentials = trustworthy content.
✅ E-E-A-T signals       - About page, Team page, years of experience, awards.
✅ Review platforms      - Trustpilot, G2, Capterra = validated by real humans.

If your site is missing most of these, your AEO score is low - even if your SEO score is excellent.

The Three Scores Your Website Actually Needs

I built this distinction into my own SEO tool at seo.devndespro.com. Every site audit now produces three separate scores:

SEO Score

Traditional search health. Title tags, meta descriptions, page speed, backlinks, canonical URLs, technical issues. What most SEO tools already measure.

✨ AI Snippet Score

On-page structure for AI extraction. FAQ schema, HowTo schema, question-based H2/H3 headings, concise answer paragraphs (40-80 words), entity clarity in the first 100 words.

AEO Score

True off-page AI citation signals. Bing indexing, Reddit presence, named author with schema, E-E-A-T signals, authoritative outbound links, review platform listings.

A site can score 90 on SEO and 42 on AEO. Both numbers matter now.

What a Low AEO Score Looks Like

Here is what a typical audit reveals for a well-ranked but AI-invisible site:

❌ CRITICAL  Not indexed on Bing - ChatGPT uses Bing to find sources
❌ CRITICAL  Not found on Trustpilot or G2 - no external validation
⚠️ WARNING   No author entity - AI cannot attribute or trust this content
⚠️ WARNING   Only 1 E-E-A-T signal - add About, Team, credentials
⚠️ WARNING   No Reddit mentions - Perplexity uses Reddit as a citation source
⚠️ WARNING   No authoritative outbound links - signals low credibility

None of these are traditional SEO problems.
None of them would appear on a standard SEO audit.
But every single one reduces your chances of being cited by an AI.

✅ How to Start Improving Your AEO Score

You do not need to rebuild your site. Most AEO fixes are content and configuration changes:

1. Submit your sitemap to Bing Webmaster Tools

Go to webmaster.bing.com, add your site, submit your sitemap. This gets you indexed on Bing and dramatically improves your visibility in ChatGPT results.

2. Add an author entity to your content

Add an author field to your JSON-LD schema with the author's name, URL, and credentials. AI engines use this to assess content credibility.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "author": {
    "@type": "Person",
    "name": "Your Name",
    "url": "https://yoursite.com/about"
  }
}

3. Build your About and Team pages

An AI reading your site needs to understand who you are and why it should trust you. A missing About page is a red flag for AI citation engines.

4. Add FAQ schema to your key pages

Wrap your Q&As in FAQPage JSON-LD. This is one of the clearest signals to AI engines that your page contains structured, citable answers.

5. Get listed on Trustpilot or G2

Even a handful of reviews on a recognized platform signals legitimacy. AI engines cross-reference these when assessing brand trustworthiness.

6. Build a Reddit presence

Comment genuinely in relevant subreddits. Over time, mentions of your brand on Reddit become citation signals for Perplexity and other AI engines.

The Shift is Already Happening

A growing percentage of informational searches now get answered by AI without any click-through to a website. That number will only increase.

The question is not whether AI search matters.

It is whether your site is ready for it.

Check Your Own AEO Score

I built the AEO audit into seo.devndespro.com because I could not find any tool that checked these signals together in one place.

Run a fresh audit and look at the AEO tab. You will likely find a gap between your SEO score and your AEO score.

That gap is your opportunity.

Mahadevan is a full-stack developer and UI/UX designer running devndespro, a web development and SEO studio based in Stavanger, Norway. He writes about modern search, AI optimization, and building tools that help businesses stay visible in an AI-first world.

Free AI SEO Checker: 39 AEO and GEO Checks Explained

Tue, 26 May 2026 14:21:23 +0000

Free AI SEO Checker: 39 AEO and GTraditional SEO tools check what Google cares about. This one checks what AI systems care about.

AutomateLab just shipped a free, no-signup checker that audits any URL across 39 checks for answer-engine optimization (AEO) and generative-engine optimization (GEO). Here's what each check actually tests.

The gap this tool fills

Most SEO auditors tell you your title is too long or your Core Web Vitals are poor. Useful for Google's ranking pipeline. But AI search engines like ChatGPT, Perplexity, and Claude extract answers differently. They look at:

Whether their crawlers can access your content
Whether your opening paragraph resolves the query directly
Whether your structured data schema type matches what the page actually says
Whether your content exists in raw HTML (not locked behind JavaScript execution)

None of that is covered by existing tools. That's the gap.

Five categories, 39 checks

Crawlability (6 checks)

The tool checks whether GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot, PerplexityBot, and Amazonbot are blocked in robots.txt. Many sites block these accidentally via wildcard disallow rules. It also checks robots.txt accessibility, sitemap presence, and canonical URL consistency.

Structured Data (7 checks)

Validates JSON-LD presence, schema type alignment (does @type: Article match what you published?), required properties per schema type, OpenGraph and Twitter cards, and author attribution. The author check looks for Wikidata or LinkedIn identity links from the author page.

On-Page Content (11 checks)

The most nuanced category. The question-heading check is based on a study showing 72% of ChatGPT-cited pages use question-style H2-H4 headings. The tool flags pages below a 30% ratio. Other checks cover reading level, answer density, internal link structure, image alt text, and heading hierarchy.

AI Readability (8 checks)

Answer density: First paragraph should resolve the implied query in under 60 words
Render mode: Page must contain 200+ visible words in raw HTML; JavaScript-rendered content is invisible to non-executing bots
llms.txt: Checks for presence and accessibility
Entity recognition: Named entity density signals
Outbound authority: Links to authoritative external sources

Technical Health (7 checks)

HTTPS, mixed content, mobile viewport meta tag, payload compression, image dimension attributes, HSTS headers, and security headers. Table-stakes checks, but the tool catches edge cases.

Scoring

Checks are weighted 1-5 based on signal strength. The tool produces category scores and a weighted overall letter grade:

A: 90+
B: 75-89
C: 60-74
D: 40-59
F: below 40

Each failed check includes the evidence string (what the tool found) and remediation guidance (what to change).

How it works

User pastes a URL
Server-side proxy fetches HTML, robots.txt, sitemap, and llms.txt in a single request
All 39 checks run client-side in the browser
Results render immediately with per-check evidence

Nothing is stored. No login required. Single URL per run.

For teams that need bulk audits, @automatelab/ai-seo-mcp wraps all 39 checks for Claude Code and Cursor workflows.

Try it

https://automatelab.tech/blog/al-products/free-ai-seo-checker-aeo-geo/

Part of the AutomateLab AI automation stack.EO Checks Explained

Citation Intelligence MCP

Mon, 25 May 2026 14:54:29 +0000

We built a free, MIT-licensed MCP server for tracking which URLs ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, and Bing cite for any query.

12 tools across three jobs:

Track citations: check_citations, am_i_cited, ai_overview, cited_for
Predict citations: predict_citation, compare_domains, wikipedia_mentions
Find leaks: gsc_citation_gap, audit_sitemap, citation_trend

Three of the tools run on a local cache and cost zero. The rest use your own API keys (Anthropic, OpenAI, Gemini, SerpAPI) at roughly $0.01 to $0.03 per query. Hosted alternatives (Profound, AthenaHQ, Otterly, Ahrefs Brand Radar) start at $295/month.

Install:

`npx -y @automatelab/citation-intelligence`

Wires into Claude Code, Claude Desktop, Cursor, n8n, or any MCP host.

Launch write-up: https://automatelab.tech/launching-the-citation-intelligence-mcp/

Feedback welcome.

AI-SEO for Developers: How to Get Your Docs and Blog Cited by ChatGPT, Claude, and Perplexity

Thu, 21 May 2026 12:37:24 +0000

Why AI-SEO is suddenly the most important SEO surface, exactly how to do it, and what it actually pays back.

Three years ago you optimized a page to rank on Google. Today the page that ranks #1 still loses six out of ten clicks the moment an AI Overview shows up above it. The new game is not ranking — it is being cited inside the answer itself. This guide is the developer-flavored playbook: why AI-SEO matters now, exactly how to do it on a site you control, and the numbers it pays back.

What is AI-SEO, AEO, GEO, and LLMO?

AI-SEO is the umbrella term for getting your pages surfaced, summarized, and cited by AI assistants. Three sub-disciplines sit underneath it:

AEO — Answer Engine Optimization. Targets the structured answers shown inside ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews.
GEO — Generative Engine Optimization. Targets the way generative models compose multi-source answers and decide which sources earn a citation.
LLMO — LLM Optimization. The broader practice of shaping content, schema, and crawler access so large language models treat your pages as the authoritative source for a topic.

In practice the three overlap heavily. The signals that earn an AI Overview citation also earn a Perplexity citation also earn a ChatGPT Search citation. So most of this guide treats them together and only splits them when the implementation diverges.

Why does AI-SEO matter now?

Four numbers explain why every team with a public site should already have an AI-SEO playbook running.

1. The audience already moved

ChatGPT now serves more than 2 billion queries per day across roughly 900 million weekly active users. Perplexity crossed 45 million monthly active users and is growing 66% year over year. These are not toy numbers — this is where buyers, developers, and decision-makers go first when they have a question your product answers.

2. AI Overviews dominate the SERP

Roughly 55 to 60 percent of US Google searches now show an AI Overview. In health, that number is 82 percent. For most informational queries, the AI answer is the first thing a user sees — and whether they keep scrolling depends almost entirely on whether your page made it into that answer.

3. Classic CTR is collapsing

Authoritas measured a 61% average drop in organic CTR for queries where an AI Overview is present — and on some verticals the top result loses 79% of its clicks. Even ranking #1 is no longer enough. The floor that classic SEO defended for twenty years is gone.

4. Cited brands win the click-through battle anyway

Here is the upside. When your brand is cited inside the AI answer, organic clicks go up 35% and paid clicks go up 91%. AI-referred traffic, when it does click through, converts at 4 to 5 times the rate of Google organic across a dataset of 12 million visits. The buyers who arrive via AI are pre-qualified by the model.

How is AI-SEO different from classic SEO?

Classic SEO optimized a page to rank. AI-SEO optimizes a page to be extracted, summarized, and cited. The work overlaps — both reward fast, well-structured, authoritative pages — but the levers diverge once you get past the basics.

Layer	Classic SEO	AI-SEO
Goal	Top 10 ranking	Be cited inside the answer
Primary signal	Keywords + backlinks	Quotable answers + entity authority
Page structure	Title + meta + Article schema	H2 = the exact question + FAQPage + HowTo
Crawler control	Allow Googlebot	Per-bot rules for 10+ AI crawlers
Discovery surface	sitemap.xml	sitemap.xml + llms.txt
Win condition	A click	A brand mention + a click
Measurement	Rank + CTR	Citation share + AI-referred conversion

The good news for anyone worried about a tradeoff: almost everything you do for AI-SEO also helps classic Google. Schema, FAQ structure, sitemap hygiene, canonical correctness, Core Web Vitals, entity links — all of these have been ranking factors for years. The only AI-specific additions are llms.txt and per-crawler robots.txt allow rules, and neither of those affects Googlebot. The work is additive, not a tradeoff.

How do you do AI-SEO? A seven-step technical playbook

This is the sequence we run on every site, in the order we run it. The first three steps are infrastructure and usually take a day. Steps four through seven are content-shaped and run for as long as you keep publishing.

Step 1. Audit and configure per-bot crawler access

There are at least ten AI crawlers worth thinking about: GPTBot, ChatGPT-User, OAI-SearchBot (all OpenAI), ClaudeBot and Claude-Web (Anthropic), PerplexityBot, Google-Extended, Bingbot for Copilot, Applebot-Extended, and Bytespider. Each has a different role. GPTBot trains ChatGPT; OAI-SearchBot powers ChatGPT Search citations; ClaudeBot powers Claude Search; Google-Extended is the gate for Gemini training (separate from Googlebot).

The trap we see most often: a site blocks GPTBot to opt out of training but leaves OAI-SearchBot allowed, which means ChatGPT can answer about your page but cannot learn from your fresh content — and the resulting citation gets stale fast. Or the inverse: a site allows GPTBot but accidentally blocks OAI-SearchBot, killing real-time citation eligibility entirely. Map the access matrix explicitly, do not inherit it.

A reasonable starting robots.txt for a site that wants both training and citation visibility:

User-agent: GPTBot
Allow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

Sitemap: https://example.com/sitemap.xml

Step 2. Ship an llms.txt

llms.txt is to LLMs what robots.txt was to search crawlers: a tiny, human-and-machine-readable file at the root of your domain that points models at the canonical entry points for your content. Place it at /llms.txt. Keep it short. A workable minimum:

# Example Inc
> One-line summary of what the company does.

## Docs
- [Quickstart](https://example.com/docs/quickstart): how to install and run in 5 minutes
- [API reference](https://example.com/docs/api): full REST and SDK reference

## Blog
- [Engineering blog](https://example.com/blog): performance, architecture, postmortems

## Optional
- [Open-source repos](https://github.com/exampleinc): MIT-licensed tools and SDKs

An llms-full.txt at the same level can expand into longer descriptions per page. Adoption is uneven — some models honor it, others do not — but the cost of shipping it is twenty minutes and it never hurts.

Step 3. Cover the schema basics, then go beyond Article

Every public page should ship JSON-LD. The bare minimum is Organization (one per domain, including a sameAs array pointing at your verified social profiles and Wikipedia entry if one exists) and BreadcrumbList. From there, pick the right page type:

FAQPage on any page with a Q&A block. This is the single highest-yield AI-SEO schema because the questions become extractable units.
HowTo on tutorials and step-by-step guides. Mark up each step. AI Overviews pull these almost verbatim.
SoftwareApplication on product pages.
Article or TechArticle on blog posts, with author and publisher correctly populated.

The entity links inside Organization.sameAs do more work than they appear to. They tell models that the brand on this page is the same entity as the LinkedIn page, the GitHub org, the Wikipedia article. That mapping is what lets a model say "according to Example Inc..." with confidence instead of fuzzing the attribution.

Step 4. Phrase every H2 as the exact user question

This one change is worth more than most of the technical work. LLMs extract answers by matching question patterns to nearby text. If your H2 is "Configuration", the model has to guess what question that answers. If your H2 is "How do you configure timeouts in the SDK?", the model has the question, the answer block underneath it, and a direct extraction path.

Rewrite your top 20 pages so each H2 is a literal question — the kind a buyer would type into ChatGPT. Put the one-paragraph answer immediately below, then expand. This pattern, sometimes called the inverted pyramid for AI, is the single largest content-side lever.

Step 5. Add comparison tables and step-by-step lists

Models love tables. A clean comparison table — your product vs an alternative, two approaches side by side, a feature matrix — gets quoted in AI Overviews and Perplexity answers far more often than the equivalent prose. Same for numbered step-by-step instructions. If your page contains a procedure, mark it up as an ordered list and consider also wrapping it in HowTo schema.

Step 6. Build the entity graph

Models treat a brand as the same entity across the web only if the entity is linked. Make sure your Organization.sameAs includes: LinkedIn, GitHub, Crunchbase, Wikipedia (if eligible), Wikidata (the Q-number is the entity glue), and your primary social profiles. Then make sure the bio on each of those profiles points back to your canonical homepage. This is the cheapest authority signal you can build.

Rank tracking does not capture AI-SEO performance. You need four numbers, baselined and tracked monthly:

Citation share — on your target query set, what percent of AI answers (ChatGPT, Perplexity, Claude, Gemini, AI Overviews) cite your domain?
AI Overview coverage rate — of your branded and non-branded target queries, what percent trigger an AI Overview at all? This is the addressable surface.
AI-referred traffic and its conversion rate compared to Google organic.
A per-page AI-SEO score — schema completeness, llms.txt presence, crawler access, H2 question phrasing, FAQ structure, entity coverage.

Are there tools that automate this?

Yes — and most of the audit and rewrite layer is automatable. Three projects worth bookmarking, all free and open-source:

AI-SEO MCP (free, MIT)

An MCP server that exposes 13 AI-SEO tools — audit, score, rewrite suggestions, schema generation, llms.txt validation, per-bot robots.txt analysis — to any MCP-compatible agent (Claude, Cursor, Continue, Windsurf, etc.). Vendor-agnostic, no API keys, no quota. Point it at a URL and it returns a structured score plus rewrite suggestions. Repo and install instructions: https://automatelab.tech/products/mcp/ai-seo/

Content Distribution MCP

Once you have an AI-SEO-shaped page, distribution multiplies its citation odds — more high-authority surfaces mentioning your brand means a higher chance any single LLM sees it during training or retrieval. This MCP auto-publishes to DEV.to, Hashnode, GitHub Discussions, Reddit, and Bluesky with idempotent state, and falls back to a browser bot for Medium, LinkedIn, and X. Repo: https://automatelab.tech/products/mcp/content-distribution-mcp/

Publishing Skills for Claude Code

If you are writing the long-tail content for your AI-SEO playbook with Claude Code, this is the matching skill bundle: blog-topic-research (finds question-shaped topics with real demand signals), ghost-blog-writer (drafts in the AI-SEO-friendly structure described above), and an SVG figure generator. Public repo: https://automatelab.tech/products/skills-and-plugins/publishing-skills/

How long does AI-SEO take to pay back?

Faster than classic SEO. Crawler access and schema changes pick up within days to two weeks — AI Overviews and ChatGPT Search both refresh on roughly weekly cycles, far faster than the months a new backlink takes to influence Google rank. Content rewrites and entity-graph work take four to eight weeks to show up in citation share. A well-shaped page on an established domain can earn its first AI Overview citation within a single week.

What if you do not have time to do this yourself?

If you would rather hand the whole build to a team that does this every day, AutomateLab runs a productized AI-SEO, AEO, and GEO service: audit, schema templates, llms.txt, per-crawler rules, page rewrites, entity-graph build, and a citation dashboard — shipped in three to four weeks. You own the repo, the configuration, and a forever license to the open-source AI-SEO MCP at the end of the build. Details and an intro call: https://automatelab.tech/services/ai-seo/

FAQ

Is AI-SEO the same as AEO or GEO?

AI-SEO is the umbrella. AEO (answer engine optimization) targets the structured answer surface. GEO (generative engine optimization) targets multi-source generative answers. LLMO is the broader practice. The signals overlap heavily, so most practitioners treat them as one workstream.

Will AI-SEO hurt my Google rankings?

No. Every signal that helps AI-SEO — schema, FAQ structure, sitemap hygiene, Core Web Vitals, entity links — also helps classic Google ranking. The only AI-specific additions are llms.txt and per-crawler allow rules, neither of which affects Googlebot.

Do I need to block GPTBot to protect my content from training?

Blocking GPTBot prevents training but does not prevent citation — OAI-SearchBot is a separate crawler that powers real-time ChatGPT Search answers. Most teams want OAI-SearchBot allowed even when they choose to block GPTBot. Decide each crawler explicitly; do not inherit defaults.

What is the minimum I should do today?

Three things, in this order: (1) audit your robots.txt for per-bot rules and add OAI-SearchBot and ClaudeBot allow lines if they are not there, (2) add FAQPage schema to your three highest-traffic pages, (3) rewrite the H2s on those pages to be literal user questions. That is roughly two hours of work and it usually moves the needle within two weeks.

How do I track whether I am being cited?

Manually: run your target queries on ChatGPT, Perplexity, Claude, Gemini, and Google with AI Overviews enabled once a week and log citations. Automatically: the AI-SEO MCP includes a brand-mention tracker that does this on a schedule. Track citation share over time, not just rank.

Bottom line

AI assistants are the new top of the funnel. Being ranked #1 no longer guarantees a click. Being cited inside the AI answer does. The good news for developers: most of the work is technical, deterministic, and automatable — schema, crawler rules, structured answers, llms.txt, entity links. Ship those, measure citation share, and treat the AI answer surface like the page-one ranking it has quietly become.

If you want the audit-and-rewrite tools, grab the free AI-SEO MCP: https://automatelab.tech/products/mcp/ai-seo/

If you want it built for you, the productized service lives at: https://automatelab.tech/services/ai-seo/

Originally written for coderlegion.com. Author: AutomateLab Editorial. Republish with attribution and link back to https://automatelab.tech/services/ai-seo/.

The Magento product description that AI can't read

Fri, 15 May 2026 07:26:22 +0000

Originally published on angeo.dev

You installed robots.txt rules for GPTBot. You generated llms.txt. You added Product schema. The AEO audit is green. ChatGPT still recommends your competitor.

The problem is not what you added. It is how Magento renders product content — and whether AI extraction systems can actually reach it.

The extraction layer problem

AI recommendation systems don't browse your store like a customer. They fetch raw HTML, parse it through an extraction layer, chunk the content into retrievable units, score those chunks against query intent, and synthesise an answer.

The extraction layer is where most Magento stores lose the race.

Content visibility for AI falls into four practical levels:

Level	State	Extraction reliability
1	Server-rendered, early in HTML	High
2	Server-rendered but hidden/collapsed	Reduced
3	In DOM only after JS execution	Low
4	Loaded async after user interaction	Very low

Googlebot handles levels 2–4 fine — it runs a full headless Chrome pipeline. Most AI crawlers do not. For AI, server-rendered HTML early in the document is what gets extracted reliably. Everything else is a gamble.

What Magento actually serves

Default Magento 2 wraps product descriptions inside a RequireJS tabs widget:

data-mage-init='{"tabs": {...}}'

In most Luma installations, the description is in the server-rendered HTML — but:

marked as inactive (collapsed tab)
positioned deep in the document after navigation, widget JSON, form keys, boilerplate
structurally ambiguous to an extraction parser

Run this to see your actual first impression on AI crawlers:

curl -s "https://yourstore.com/your-product-url.html" \
  | python3 -c "
import sys, re
html = sys.stdin.read()
text = re.sub(r'<[^>]+>', ' ', html)
text = re.sub(r'\s+', ' ', text).strip()
print(text[:3000])
"

If your product description text is not clearly readable in those first 3000 characters — you are at Level 2 at best, and competing at a structural disadvantage against every store that renders their description at Level 1.

Fix 1 — Activate the tab by default (quick win)

<!-- catalog_product_view.xml in your theme -->
<referenceBlock name="product.info.details">
    <arguments>
        <argument name="config" xsi:type="array">
            <item name="settings" xsi:type="array">
                <item name="active" xsi:type="boolean">true</item>
            </item>
        </argument>
    </arguments>
</referenceBlock>

Keeps the tab UI for visitors. Removes the CSS-hidden state so the description is exposed at page load. Level 2 collapsed → Level 2 visible. Takes 5 minutes, no module required.

Fix 2 — Render description above tabs (structural fix)

<!-- catalog_product_view.xml -->
<referenceContainer name="product.info.main">
    <block class="Magento\Catalog\Block\Product\View\Description"
           name="product.description.above.tabs"
           template="YourVendor_Theme::product/description-visible.phtml"
           after="product.info.price">
</referenceContainer>

<?php
// description-visible.phtml
$description = $block->getProduct()->getData('description');
if (!$description) return;
?>
<section class="product-description-aeo" itemprop="description">
    <?= /* @noEscape */ $description ?>
</section>

Description is now at Level 1 — unconditional, server-rendered, early in the document. Keep the tabs further down for UX. Both coexist without conflict.

Fix 3 — Render all tab content server-side (comprehensive)

Override details.phtml so all tab content is in the initial HTML response. The tabs widget handles visibility via CSS only — active tab is visible, others are hidden. AI parsers reading raw HTML see everything.

Hyva Theme note

Hyva uses Alpine.js x-show instead of RequireJS tabs. Description is usually in the HTML source — but Alpine bindings add structural noise. Most reliable fix: render description once outside the Alpine component, as a clean server-rendered block, early in the layout. Same pattern as Fix 2 above, same result.

Check your JSON-LD delivery

Same principle applies to schema. If JSON-LD is injected via GTM or hydrated client-side, AI parsers often miss it entirely.

curl -s "https://yourstore.com/your-product.html" \
  | grep -o '"@type": "Product"'

No output = client-side schema = invisible to most AI extraction pipelines.

Server-rendered JSON-LD in <head> is the only reliable delivery method for AI systems.

Bottom line

AEO guides focus on what to add. This problem is about how content is exposed — position in the document, server vs client rendering, structural clarity for extraction parsers.

Most Magento stores have the right content. It just never reaches the layer that determines whether their products become AI recommendation candidates.

The fix takes under an hour. No new modules, no API keys — layout XML overrides in your theme.

Run the curl test on your top-selling product now.

Full article (all three fix approaches with complete code, Hyva deep-dive, schema delivery comparison): angeo.dev →

Free AEO audit tool (CLI, MIT licensed): angeo/module-aeo-audit

Why do some pages get impressions but almost no clicks?

Mon, 04 May 2026 20:48:19 +0000

A lot of website owners open Google Search Console, see impressions going up, and assume traffic should follow.

But it often does not.

The page appears in search results, Google counts impressions, yet users still do not click. That does not always mean the page is “bad.” It usually means one of a few things is happening:

The page is ranking too low to be noticed.

The title does not match what the searcher actually wants.

The snippet looks generic compared with competing results.

The query is informational, but the page sounds too commercial.

Or the search result is being answered directly by SERP features, snippets, maps, videos, or AI Overviews before the user reaches organic results.

This is one of the most misleading SEO situations because impressions can make it look like visibility is improving, while real traffic stays flat.

The first thing I usually check is not the title tag. I check position.

If a page has impressions but sits around position 40, 60, or 80, rewriting the meta description will not fix much. That page needs stronger content, better internal links, clearer topical relevance, and probably more authority.

But if the page is already appearing around positions 5–20, then CTR work can make a real difference. That is where better titles, clearer descriptions, stronger page intent, and more useful answer-style sections can help.

I wrote a deeper breakdown here:

https://visrank.org/blog/why-my-website-has-impressions-but-no-clicks

The main takeaway: do not treat “impressions but no clicks” as one problem.

Separate it into:

Ranking-depth problem
Snippet problem
Search intent problem
Trust problem
Zero-click / AI answer problem

Each one needs a different fix.

Curious how others handle this: when you see high impressions and low clicks in Search Console, do you usually start by rewriting titles, improving the page content, or checking the SERP manually first?

AEO for Magento 2: ChatGPT Visibility Audit Module and AI Engine Optimization Checklist

Sat, 02 May 2026 21:36:35 +0000

AEO for Magento 2: ChatGPT Visibility Audit Module and AI Engine Optimization Checklist

Tags: #magento #aeo #chatgpt #ecommerce #php

Most Magento store owners focus on Google SEO. But in 2026, a growing share of purchase decisions starts in ChatGPT, Perplexity, or Gemini — not in a search bar. When someone asks ChatGPT "best wireless headphones under $100", your store either appears in the answer or it doesn't. And the reasons it doesn't have almost nothing to do with traditional SEO.

This post covers AEO (AI Engine Optimization) specifically for Magento 2 — what it is, the 9 technical signals that determine your store's AI visibility, and a free open-source audit module that checks all of them in one CLI command.

What is AEO and why Magento stores fail it by default

AEO is the practice of making your store's content accessible and understandable to AI engines — ChatGPT, Perplexity, Gemini, and similar platforms that answer purchase queries directly instead of showing a list of links.

The difference from SEO is fundamental. Google ranks your pages. AI engines decide whether to recommend your store at all. To make that recommendation, they need to:

Crawl your store (many Magento stores block AI crawlers in robots.txt)
Read structured product data (most Magento stores have incomplete Product schema)
Find a machine-readable catalog (most stores have no llms.txt or llms.jsonl)
Access a product feed (required specifically for ChatGPT Shopping results)

A default Magento 2 installation fails every single one of these. The good news: all of them are fixable in under two hours with the right modules.

The 9 AEO signals for Magento 2

Running bin/magento angeo:aeo:audit checks these 9 signals and returns a weighted score from 0–100%:

Signal #1 — robots.txt AI bot access (weight: 20%)

The most critical signal. Magento's default robots.txt typically contains Disallow: / for all bots, which blocks OAI-SearchBot (ChatGPT), PerplexityBot, Google-Extended (Gemini), and ClaudeBot.

Check your store right now:

curl https://yourstore.com/robots.txt | grep -E "OAI-SearchBot|PerplexityBot|Google-Extended"

No output means these bots are being blocked. Fix:

composer require angeo/module-robots-txt-aeo
bin/magento setup:upgrade && bin/magento cache:flush

Important distinction: GPTBot is OpenAI's training data crawler — blocking it opts your content out of model training but does NOT affect ChatGPT search or Shopping visibility. OAI-SearchBot is the live search crawler — that one must be allowed.

Signal #2 — llms.txt (weight: 18%)

llms.txt is an emerging standard (similar to robots.txt) that tells AI engines what your store is about, what products you sell, and which URLs are most important to crawl. Perplexity has specifically documented it as a crawl signal.

Check if yours exists:

curl -s -o /dev/null -w "%{http_code}" https://yourstore.com/llms.txt

404 means it doesn't exist. Fix:

composer require angeo/module-llms-txt
bin/magento setup:upgrade && bin/magento cache:flush

The module generates both llms.txt (human-readable summary) and llms.jsonl (machine-readable catalog with one JSON object per product line) via cron.

Signal #3 — Product JSON-LD schema (weight: 15%)

AI engines read your product data from <script type="application/ld+json"> blocks on product pages. Magento's default schema outputs a Product node but omits offers.availability — which is required for ChatGPT Shopping feed validation.

Check what your store outputs:

curl -s https://yourstore.com/sample-product.html | \
  python3 -c "
import sys, json, re
body = sys.stdin.read()
blocks = re.findall(r'<script[^>]+type=\"application/ld\+json\"[^>]*>(.*?)</script>', body, re.DOTALL)
for b in blocks:
    try:
        d = json.loads(b)
        nodes = [d] if isinstance(d, dict) else d
        for n in nodes:
            if n.get('@type') == 'Product':
                offers = n.get('offers', {})
                print('availability:', offers.get('availability', 'MISSING'))
    except: pass
"

If you see availability: MISSING or availability: InStock (wrong format — needs full URI), fix with:

composer require angeo/module-rich-data
bin/magento setup:upgrade && bin/magento cache:flush

The correct format is "availability": "https://schema.org/InStock" — the full URI. AI engines silently reject the short string format.

Signal #4 — AI product feed (weight: 20% combined with Signal #1)

ChatGPT Shopping product cards — the feature that shows prices, availability, and buy links in chat — requires a separate product feed submitted to OpenAI's merchant program. This is not the same as organic crawl visibility.

The feed format is ACP (Agentic Commerce Protocol) — a .jsonl.gz file with one product per line, regenerated every 15 minutes:

composer require angeo/module-openai-product-feed angeo/module-openai-product-feed-api
bin/magento setup:upgrade && bin/magento cache:flush

After installing, apply at chatgpt.com/merchants (currently US-only, waitlisted).

Signals #5–9 — sitemap, Open Graph, FAQPage schema, llms.jsonl, canonical

These are lower-weight signals but collectively add up:

sitemap.xml (7%) — must exist, have <loc> tags, and be referenced in robots.txt
Open Graph tags (10%) — og:title, og:image, og:description on product pages
FAQPage schema (5%) — on CMS pages with Q&A content
llms.jsonl (7%) — machine-readable product catalog for AI pipelines
Canonical tags (5%) — deduplication signal for AI crawlers

Running the audit

Install the audit module:

composer require angeo/module-aeo-audit
bin/magento setup:upgrade && bin/magento cache:flush

Run it:

bin/magento angeo:aeo:audit

Sample output for a typical store before any fixes:

  AEO Score: [████░░░░░░░░░░░░░░░░] 21% — Needs Improvement

  ✗ Fail  robots.txt — OAI-SearchBot          weight: 20%  → composer require angeo/module-robots-txt-aeo
  ✗ Fail  llms.txt — AI content map           weight: 18%  → composer require angeo/module-llms-txt
  ✗ Fail  Product schema — JSON-LD            weight: 15%  → composer require angeo/module-rich-data
  ✗ Fail  AI product feed                     weight: 20%  → composer require angeo/module-openai-product-feed
  ✓ Pass  sitemap.xml                         weight:  7%
  ⚠ Warn  Open Graph tags                     weight: 10%
  ✗ Fail  llms.jsonl — machine-readable       weight:  8%  → composer require angeo/module-llms-txt
  ✓ Pass  Canonical tags                      weight:  5%
  ⚠ Warn  FAQPage schema                      weight:  5%

After applying the fixes, the same store typically scores 78–87%.

The module also includes:

--format=json output for CI pipelines
--fail-on=80 to fail builds below a threshold
Admin UI under Marketing → Angeo AEO → Score Trend with historical chart
Weekly cron audit with score tracking

How each AI engine discovers your products

The three major AI platforms use fundamentally different discovery mechanisms:

Perplexity is entirely crawl-driven. Fix robots.txt, add llms.txt, fix Product schema — Perplexity re-indexes within 1–3 weeks. No merchant registration required.

ChatGPT runs on two tracks: organic crawl via OAI-SearchBot (for general query answers) and the ACP product feed (for Shopping product cards). Both are needed for full visibility.

Gemini primarily reads from Google Merchant Center. If you're already in GMC, Gemini has access almost immediately. The Google-Extended bot affects AI Overviews (non-shopping queries) but not GMC-sourced product results.

This means: Perplexity visibility is the fastest win (weeks), Gemini depends on GMC status, ChatGPT Shopping has the longest timeline (merchant approval) but the highest commercial intent.

Adobe Commerce Cloud — one extra step

On Adobe Commerce Cloud with Fastly CDN, every change to robots.txt or llms.txt requires a Fastly cache purge before AI crawlers see the new content. This is the most common reason fixes "don't work" on ACC:

php bin/magento fastly:purge:url https://yourstore.com/robots.txt
php bin/magento fastly:purge:url https://yourstore.com/llms.txt

Then verify with curl directly — not the admin config — before considering the fix complete.

The full checklist

Quick reference for the complete implementation:

# 1. Diagnostic — run first
composer require angeo/module-aeo-audit
bin/magento setup:upgrade && bin/magento cache:flush
bin/magento angeo:aeo:audit

# 2. Fix robots.txt
composer require angeo/module-robots-txt-aeo

# 3. Generate llms.txt + llms.jsonl
composer require angeo/module-llms-txt

# 4. Fix Product schema
composer require angeo/module-rich-data

# 5. AI product feed (ChatGPT Shopping)
composer require angeo/module-openai-product-feed angeo/module-openai-product-feed-api

# Apply all at once
bin/magento setup:upgrade && bin/magento cache:flush

# Verify
bin/magento angeo:aeo:audit
curl https://yourstore.com/robots.txt | grep OAI-SearchBot
curl https://yourstore.com/llms.txt | head -3

Packagist

All modules are MIT licensed, free, and available on Packagist:

angeo/module-aeo-audit — audit tool
angeo/module-robots-txt-aeo — robots.txt fix
angeo/module-llms-txt — llms.txt + llms.jsonl
angeo/module-rich-data — Product schema
angeo/module-openai-product-feed — ACP feed
angeo/module-openai-product-feed-api — REST API
angeo/module-openai-instant-checkout — instant checkout from ChatGPT

Source: github.com/angeo-dev/module-aeo-audit

Full guide with deeper dives on each signal: angeo.dev/magento-2-aeo-guide

Originally published at angeo.dev

Coder Legion - Recent posts tagged aeo

Two thirds of what Google's AI Overviews cite comes from the top 30% of a page.

We ran an AI SEO audit on 50 SaaS sites, here are the key findings

From SEO to AI Search: Jarno S’s Journey Into Making the Web More Understandable to AI

From SEO to AI Search: Jarno S’s Journey Into Making the Web More Understandable to AI

From Traditional Search to AI Search

The Question That Changed Everything

AI Visibility Should Be Measured

Think of AI Visibility Like Regression Testing

A Mention Isn't the Same as a Citation

Structured Data Helps AI Understand the Pieces

What About llms.txt?

Make Pages Easy to Quote

The AI Search Advice He Thinks Is Overhyped

What AI Search Reveals About Businesses

Building AEOvara

From AI Visibility to Repeatable Benchmarks

Lessons From Photography

Where Search Goes Next

Advice for Developers and Small Businesses

A Lesson From Building a New Category

AI Search Is an Information-Quality Challenge

What's Next for AEOvara?

About Jarno S

CoderLegion Developer Stories

Your AI Visibility Score May Be Wrong: A Reproducible Testing Protocol

First, identify what actually changed

1. The brand changed

2. The engine changed

3. The sample changed

A prompt is not a keyword

Use prompt families, not a bag of convenient questions

Record the test environment like an experiment

Measure a rate—and show its uncertainty

Do not use one noise budget for every metric

Establish a noise floor before claiming a win

Build a holdout set to catch overfitting

Publish the method, including its limitations

A minimum viable reproducibility checklist

The bottom line

FAQ

How many times should I run each AI-search prompt?

Should I freeze my prompts or update them?

Can I compare an API result with the consumer AI product?

Is a higher mention rate proof that my content change worked?

Does a citation mean the AI answer used my page correctly?

Sources and further reading

AEO for Finnish SMEs: A Practical 2026 Guide to Getting Cited by AI

What is Answer Engine Optimization?

Why Finnish SMEs need AEO in 2026

The shift from rankings to answers

Start with real customer questions

Build topic authority, not a pile of random posts

Use structured data with purpose

Win the local answer

Publish content formats that answer buying questions

FAQ blocks

Comparison pages

Pricing pages

Step-by-step guides

Think multilingual from the start

Get the technical foundations right

Demonstrate experience and trust

A practical 30-day AEO plan

Week 1: Find the gaps

Week 2: Fix the technical layer

Week 3: Strengthen commercial content

Week 4: Expand and measure

Common AEO mistakes

The takeaway

Frequently asked questions

Is AEO replacing SEO completely?

Can a small Finnish business compete with larger brands?

How quickly can AEO produce results?

Does every Finnish SME need multilingual content?

What is the best first step?

About the author

AI Search Visibility Is Finally Measurable: A Developer’s Field Guide for 2026

The short answer

Why the old SEO dashboard is not enough

What About `llms.txt`?