AEO: SEO Isn’t Dead, but the Rules of Visibility Have Changed

TLDR

TLDR People no longer search only on Google: they ask ChatGPT, Gemini, and Perplexity. AEO, Answer Engine Optimization, is the work of making a company understandable and citable by these systems. It does not replace SEO; it is the new layer of visibility. This article covers what the data says, what actually works, and what is just vendor marketing.

Introduction

When you ask a new customer how they found you, one answer is becoming more and more common: ChatGPT suggested you. For twenty years, marketers had a single question, how to reach the first page of Google. That question has not disappeared, but on its own it is no longer enough.

People no longer search with just two or three words. They open ChatGPT, Gemini, Perplexity, or Copilot and describe a whole situation, expecting an answer, a comparison, sometimes a direct recommendation. This changes what visibility means. For years it meant appearing in a list of links; now it means being the content that an AI system reads, interprets, and decides to cite or mention.

Saying SEO is dead is wrong: Google remains dominant. But any B2B company that has never checked what ChatGPT or Gemini say about it is already behind. The right question is not SEO versus AEO. It is this: is AEO the next technical phase of SEO, or a broader visibility strategy for a buying journey that now also runs through AI? This article argues for the second answer.

What Is AEO?

AEO, Answer Engine Optimization, is the set of choices that make a company understandable and extractable by the systems that generate answers. A cluster of related acronyms has grown around the shift, and they are worth separating.

AEO. The structural layer: how content is formatted, with question-based headings, FAQ sections, structured data, and short paragraphs.

GEO. Generative Engine Optimization, the academic term, from a study by Princeton University researchers: depth, authority, and density of data, so the model trusts the content, not only reads it.

LLMO. Large Language Model Optimization, the entity-focused approach: how a model categorizes a brand and which market it associates with it.

AI visibility. Not a discipline but the metric: how often, and with what sentiment, a brand is cited, mentioned, and recommended.

For an SME the distinction matters little: these are different layers of the same shift, and they work together. This article uses AEO as the umbrella term.

Why Does AEO Matter Now?

The urgency does not come from vendor marketing, but from measurable shifts.

Zero-click search. The SparkToro study of 2024 data finds that 58.5% of Google searches in the US and 59.7% in the EU end without a click to an external site. Solid, independent data, but it covers only Google, only the US and Europe, and counts traffic to Google-owned properties as a non-click.

Google AI Overviews. Ahrefs’ analysis of 300,000 keywords links the presence of an AI Overview to a 58% drop in clicks for the top-ranking page, in line with other studies showing declines between 45% and 65%. This is a correlation, and Ahrefs is an SEO vendor.

AI platform adoption. Figures on user numbers and growth should be treated with caution: aggregator sources, opaque methods. The direction is credible, the precise numbers far less so.

B2B buying behavior. A Gartner survey of 645 buyers (Aug-Sep 2025) finds that 67% prefer to buy without a sales rep and 45% used AI during a recent purchase. The same survey adds a figure that reframes the picture: 69% turn back to sales reps to validate AI-generated insights. If AI describes you poorly or does not name you, you lose the research phase, but validation stays human.

In short: traditional search has not been replaced, but a growing share of discovery now happens inside AI answers, with very uneven exposure across sectors.

Is SEO Dead?

No. But it helps to distinguish five ways a company can be visible today: classic ranking on Google, a cited link inside an AI answer, an unlinked mention, a recommendation in a comparison, and use as the source of a fact. Traditional SEO mostly optimized the first; AEO works on all five. The table summarizes the operational differences.

Dimension	Traditional SEO	AEO / AI visibility	Strategic implication
Query type	Short keywords, 2-4 terms	Complete, contextual questions, often 7 or more words	Map buyers’ real questions, not just search volumes
Ranking signal	Backlinks and domain authority	Entity clarity, data density, structured data, external consensus	Authority is built with proof and consistency, not links alone
Content format	Long narrative text, to keep the reader on page	Modular blocks, short fact-first paragraphs, lists and tables	Content must be made easy for a machine to extract
Measurement	Position, impressions, organic traffic	Citation frequency, share of voice, sentiment, mentions	New tools are needed, and you measure a trend, not a fixed number
Conversion path	Click to the site, then navigation	Sometimes no click: the decision forms inside the AI answer	The key message must be clear in the source, not only on the site
Trust signal	Quantity and quality of inbound links	Source citations, proprietary statistics, schema, third-party reviews	Trust becomes a mosaic of distributed proof, not one indicator
User journey	The user compares links and synthesizes alone	The AI synthesizes sources and delivers a single answer	You lose control over the order in which information is presented
Competitive set	Sites ranking on the same keyword	The brands the AI places next to you in a comparison	The model picks your competitors, and they may not be the expected ones
Content refresh cycle	Periodic update, useful but not urgent	Frequent update: freshness weighs in source selection	You need a continuous update process, not an annual review
Role of brand or entity	Important but secondary to keywords	Central: the AI must understand who you are before recommending you	Company identity becomes a technical asset, not just branding

The logic shifts: in SEO a backlink is a vote of confidence; in AEO what matters more is whatever reduces risk for the model, a verifiable statistic or valid schema. Backlinks do not become useless, but the center of gravity moves.

What HubSpot Gets Right, and How Well It Holds Up

One source for this article is the transcript of a HubSpot talk by Asia Frost. Useful material, but not neutral: in the same talk HubSpot launches its own monitoring software. Its seven ideas, in brief.

1. The blog is not dead. Supported: AI heavily cites structured content. But the 62.1% citation share HubSpot reports is proprietary data; other analyses put it at 45-63%.
2. From keywords to context. Supported: queries are getting longer, and long queries trigger an AI Overview more often. The 350-word average conversation figure has no clear source.
3. Structure matters. Plausible, but this is correlation, not cause. Schema markup rests on solid ground; other claims, such as having more than 20 external links, are fragile.
4. Freshness matters. Supported: content cited by AI is on average more recent. Updating prices and products pays off, as long as you update the data and not just the date.
5. Backlinks do not predict AI visibility. Plausible but to be handled with care: one example is not proof, and links alone are no longer enough, without being useless either.
6. AI looks for consensus on LinkedIn, Reddit, and YouTube. Supported, but different models cite different sources: optimizing for an average AI is a mistake.
7. AI results are volatile. Plausible, but with a conflict of interest: extreme volatility sells monitoring tools, and HubSpot has just launched one.

In brief: HubSpot captures the mechanics of AEO, but stays quiet on the downside, namely difficult measurement, legal risk, and partisan data.

What Companies Get Wrong About AEO

The most frequent mistakes, before the recommendations:

• Treating AEO as a trick or a plugin: there is no setting to switch on.
• Filling the blog with generic AI-generated content: it does not stand out, so it does not get cited.
• Optimizing for keywords instead of buyers’ real questions.
• Leaving company identity vague: if you do not say who you are and what you do, the AI guesses, and often gets it wrong.
• Abstract product pages, full of slogans and short on concrete facts.
• Hiding prices and comparisons, exactly what buyers ask AI for.
• No original data or case studies: without proof there is nothing to cite.
• No credible presence beyond your own site.
• Publish and forget, with no update process.
• Measuring AI visibility with occasional checks instead of as a trend.

What AI Needs to Understand About Your Company

Before any tactic comes clarity. Make it easy, for a machine and for a person, to understand:

• who you are: name and category;
• what you do, concretely and not in slogans;
• who you serve and which use cases you are best for;
• which problems you solve and where you operate;
• why you are credible, with real expertise and proof;
• how you compare to the alternatives;
• which questions buyers ask before choosing you.

If this information does not exist in explicit form, no AEO tactic works: you would be optimizing a void.

An Eight-Step AEO Framework

1. Map buyer prompts. Gather the real, contextual questions from sales and support, not search volumes.
2. Audit current AI visibility. Run 30-50 prompts across the main AI engines, noting whether and how they name you and who appears instead.
3. Clarify entity and positioning. State category, market, and value proposition, with consistent structured data.
4. Build answer-ready content. Question-based headings, a summary up front, short paragraphs that open with the fact.
5. Create persona and use-case pages. Move from generic keywords to concrete, specific situations.
6. Add proof. Case studies, proprietary statistics, benchmarks: this is what makes you citable.
7. Build external consensus. A credible presence on LinkedIn, YouTube, reviews, and industry sources, without spam.
8. Measure trends, not snapshots. Track mentions and citations over time, not a single day’s snapshot.

On the tools side, search intelligence platforms built for SEO are moving onto this ground: Pi Datametrics, for example, has added AI visibility tracking alongside traditional SEO monitoring. The usual caveat for any tool applies: it serves after the content work, not instead of it. Disclosure: the author of this article worked at Pi Datametrics between 2016 and 2018, and cites it from direct knowledge of the category, not from any current commercial relationship.

Which AEO Tactics Actually Work

The table sorts tactics by real evidence level. Where the level is low, the tactic may make sense but is not proven: it is a bet, not a certainty.

Tactic	Why it may work	Evidence level	Best use case	Risk or limitation
Question-based headings	They mirror the structure of the query, easy to extract	Medium	Informational articles, FAQs, guides	Observed correlation, not proven cause
FAQ sections	They pre-package question-answer pairs in the format AI looks for	Medium-high	Product, pricing, support pages	Useless if the answers are generic
FAQ schema and structured data	They remove ambiguity about the entity, backed by Google documentation	High	Pricing, product, FAQ pages	Needs technical skill; syntax errors break it
Clear service and product pages	They give the machine concrete facts to extract	High	All companies	None significant, it is a baseline requirement
Persona and use-case pages	They answer specific, contextual queries	Medium-high	B2B with distinct segments	Production cost, risk of overlap
Comparison pages	They intercept the buyer’s decision phase	Medium	Markets with defined competition	The comparison must stay honest and current
Pricing pages	Buyers ask AI for prices, and AI repeats them	Medium-high	Companies that can publish prices	Publishing prices is a commercial choice, not just SEO
Original data and research	They give the model unique, verifiable facts to cite	High	Industry benchmarks, surveys, internal data	Cost and time to produce valid data
Case studies	They turn client results into citable proof	High	B2B and professional services	They require client consent
Expert quotes	They add authority and a human voice	Medium	Analysis and opinion content	They must be real, not decorative
Visible last-updated date	It signals freshness, which weighs in source selection	Medium	Content on fast-changing topics	Useless if the content is not actually updated
YouTube videos with transcripts	YouTube is heavily cited; a transcript makes the video readable	Medium	Demos, explainers, customer cases	Requires steady video production
LinkedIn thought leadership	LinkedIn is cited for professional questions	Medium	Executives and reps with real expertise	Works only with authentic content
Digital PR	It builds mentions on authoritative third-party sources	Medium	Companies with news or data to share	Cost and time, results not guaranteed
Third-party reviews	They provide the external consensus AI looks for	Medium-high	Software, services, e-commerce	Fake or inflated reviews can backfire
Wikipedia or Wikidata presence	It strengthens the entity in the models’ knowledge bases	Medium	Companies with genuine encyclopedic relevance	Wikipedia has its own criteria; you cannot buy an entry
Technical crawlability	If bots cannot read the site, nothing else counts	High	All companies	A prerequisite, not a competitive advantage
llms.txt file	It would give AI crawlers a concise map of the site	Low	Technical documentation, SaaS	A proposed standard, not adopted by the main engines

A note on the llms.txt file. It gets a lot of attention, but today it is a proposal, not an adopted standard: there is no sign that the main AI engines use it. It costs little to implement, but treating it as a priority, or paying to optimize it, is not justified by the evidence.

The 30-Day AEO Plan

For a company that already has a site and a blog. Week 1: gather 30-50 real buyer questions and run them through the AI engines, setting the baseline. Week 2: update five strategic pages with question-based headings, a summary up front, FAQs, schema, a visible update date, and one original data point. Week 3: create two to five use-case or comparison pages. Week 4: add proof, fix schema and internal links, start distribution on one external platform, and define how you will measure. Thirty days are not enough to win at AEO: they are enough to stop flying blind and start a process.

Risks and Limitations

An honest picture requires naming the risks. AI answers are unstable and citations inconsistent: appearing once guarantees nothing. AI makes mistakes and can attribute wrong information to you. Measurement is imperfect: traffic from ChatGPT or Perplexity is often classified as direct. Not all sectors are equally exposed, and Google remains the main channel. Part of your AI visibility depends on what others say about you. Over-optimization, such as inflated reviews, can backfire. Finally, there is a systemic risk: legal cases between publishers and AI platforms could change how engines cite sources. Invest in what holds value either way: clear content and real proof.

Conclusion

Is AEO the next phase of SEO or a broader strategy? The second, but without drama: it does not replace SEO, it sits alongside it as the new layer of visibility. It is not a technical capability to delegate to a tool, it is an organizational capability: knowing who you are, saying it clearly, proving it, and keeping it current.

Almost everything that works for answer engines works because it also works for people. The point is not to appear once. It is that, when a customer looks for you by asking an AI, the answer is there, it is accurate, and it is yours.

Frequently Asked Questions

Does AEO replace SEO?

No. Google remains the main channel in most sectors and SEO still drives traffic. AEO is added on top: it governs visibility inside AI-generated answers.

Does every company need AEO?

Not to the same degree. A B2B or digital company is highly exposed; a local business much less so, at least for now. The investment should be sized to the sector and the type of customer.

How much does AEO cost?

It does not require an enterprise budget. It requires discipline and consistency: clear content, real proof, constant updating. It is more a question of method than of spend.

Where do you start?

With an audit. Gather 30-50 real buyer questions and run them through ChatGPT, Gemini, and Perplexity: find out whether they name you, how, and who appears instead. That is your baseline.

Final Checklist

• Can AI understand who we are and what category we operate in?
• Can AI understand who we serve and for which use cases?
• Is it clear why we are credible, with real expertise and proof?
• Do we answer buyers’ real questions, not just keywords?
• Is our content current and well structured?
• Do we have original proof: data, case studies, verifiable numbers?
• Do we have a presence beyond our own site?
• Are we tracking mentions and citations as a trend over time?

About the Author

Robert Julian Smith is an AI trainer for SMEs and a marketing consultant, based in Umbria, Italy. Before moving into training, he spent around eight years in the search software industry: from 2011 to 2015 as European Sales Director at Intelligent Positioning, a London-based SEO software house, and then until 2018 in international business development at Pi Datametrics, a search intelligence platform. He came to know SEO from inside the industry that builds it, well before anyone spoke of AEO.

Today he trains marketing and sales teams at SMEs on the practical use of AI, and has been a guest lecturer at Luiss Guido Carli University since 2018. This article comes from that double perspective: search seen from the side of the industry that builds the tools, and AI seen from the side of companies’ real processes.

Sources

The verified statistics cited in this article come from the primary sources listed below.

• SparkToro, 2024 Zero-Click Search Study (2024 data, US and EU).
• Ahrefs, AI Overviews Reduce Clicks by 58% (analysis of 300,000 keywords, Dec 2023 vs Dec 2025).
• Gartner, 67% of B2B Buyers Prefer a Rep-Free Experience (survey of 645 buyers, Aug-Sep 2025).
• Gartner, 69% of B2B Buyers Turn to Sales Reps to Validate AI-Generated Insights (same survey, presented May 2026).

Aggarwal et al., GEO: Generative Engine Optimization, Princeton University and collaborators, arXiv 2311.09735.