Home / AEO Guide

Answer Engine Optimization (AEO):
A Field Guide

AEO is sold as a brand-new discipline with its own rules, tools and agencies, while Google’s guide to its AI features lists AEO and GEO as terms in use and treats the work as SEO. The difference matters for budgets. A separate discipline implies a new budget line, a new vendor and possibly a new hire, while SEO with a few new outcomes means pointing an existing team at different targets. Each section of this guide weighs in on that question, and the guide ends with an answer on how much of the work is new.

This field guide is for marketing leads, in-house SEOs, founders and agency strategists who need to decide what to change and what to ignore. Anyone already doing SEO will recognize much of the work. Crawling, content and links make up most of the overlap between AEO and SEO, which leaves fewer genuinely new tasks than the terminology suggests.

Every claim names its source, so you can see what the AI vendors document, what independent studies found and what is still a practitioner view.

0%of AI Overview citations also rank top 10Ahrefs, March 2026
0M+ChatGPT weekly active usersOpenAI, February 2026
0.0%of referrals come from AI chatbotsStatcounter, September 2026

// 01 — The Definition

What Is Answer Engine Optimization (AEO)?

Answer engine optimization is SEO work with one extra outcome in mind, which is getting your pages and brand named, linked and described correctly inside AI-written answers.

Definition

Answer engine optimization (AEO) is the practice of making your content and brand easy for AI answer engines, such as Google AI Overviews, ChatGPT and Perplexity, to find, understand and cite. It builds on SEO and adds a new goal of appearing inside the answer itself, as well as in the results below it.

SEO aims to rank your pages in search results, and AEO aims to get your pages and brand into the answers AI systems write. Both rest on the same work, which means pages crawlers can reach, content worth trusting and a reputation other sites confirm.

The term is older than ChatGPT. Marketers first used “answer engine” for featured snippets and voice assistants, which give one answer instead of a list. Generative AI widened the idea, since answers now arrive as written summaries with several cited sources.

Is AEO Just SEO?

Google’s position is that it is. Its guide to optimizing for generative AI features, published in May 2026, lists AEO and GEO as terms people use and treats the work as SEO.

Independent data agrees only in part. Ahrefs found that only 38% of URLs cited in AI Overviews also ranked in the top 10 for the same query (863,000 SERPs, March 2026), down from about 76% in July 2025. Ranking still helps, but it no longer guarantees a citation, and Google’s statement also says nothing about ChatGPT, Perplexity or Claude, which run their own crawlers and indexes.

Our house line follows from both. AEO is the part of SEO that decides whether AI answers cite you, and most of it is the SEO you already know.

What AEO Adds to SEO

The new parts are smaller than vendors suggest. They cover crawler choices for engines beyond Google, prompt-based measurement and checking how answers describe you. The sections below check each area against that list and show where AEO is plain SEO, where it bends and where it adds work.

What AEO adds is a set of outcomes SEO never measured. Classic SEO reports rankings, clicks and conversions, and AEO adds three more questions.

01

Are you cited?

An AI answer links to your page as a source.

02

Are you mentioned?

The answer names your brand, with or without a link.

03

Are you described correctly?

The answer gets your product, pricing and positioning right.

None of these show up in a rank tracker, which is why measurement gets its own section later in this guide.

// 02 — The Vocabulary

AEO, GEO, LLMO and the Other Names

The vocabulary of AEO is a set of competing names for the same goal of getting a brand seen, cited and described accurately inside AI-generated answers.

Google lists AEO and GEO as terms in use and treats both as SEO. Nobody owns these names, so vendors, agencies and analysts use whichever suits their audience.

You’ll meet six names often, each with its own users and its own emphasis.

TermWho uses itWhat it points at
AEO (answer engine optimization)SEO agencies, marketing teams and software vendorsBeing the direct answer or a cited source in AI answers and answer features
GEO (generative engine optimization)Researchers and AI visibility vendorsVisibility inside answers written by generative AI
GEO SEOSearchers mixing both termsThe same goal as GEO, often confused with geographic or local SEO
LLMO and LLM SEOTechnical SEOs and some tool makersVisibility in large language model answers, including ones built without live search
AI SEO and AI search optimizationMarketers and publishersA broad label for any SEO work aimed at AI search
Search everywhere optimizationSome agenciesVisibility across search engines, AI tools, social platforms and marketplaces

These groupings are our editorial read of how the terms are used, not formal definitions. For you, the practical point is that six names describe one job, so a vendor’s choice of label tells you nothing about what it will do differently.

Search demand shows which names people actually type. In October 2026, Ahrefs put US monthly searches at about 7,000 for “answer engine optimization”, 6,500 for “generative engine optimization” and 800 for “llmo”. AEO and GEO are close, and LLMO is far behind, though one month of data says nothing about direction.

The names share a foundation and differ mainly in the outcome they stress, which explains the confusion around AEO terminology. Teams can lose whole meetings to the label question, so pick one name, define it in a sentence and move on to the work.

// 03 — The Business Case

What AI Search Is Worth (The Business Case in Ranges)

The business case for AEO is an estimate of what AI visibility is worth to your business, and today the honest answer is a range of figures.

AI answers change how people click, while the traffic they send directly is still small. Four findings frame the case, each listed here with its source, date and caution.

FindingFigureSource and dateCaution
Fewer clicks when an AI summary appears8% of visits with an AI summary included a click on a regular result, against 15% without onePew Research Center, March 2025 browsing data, 900 US adults, 68,879 searchesOne month of US data; 18% of searches in the sample produced a summary
Top results lose clicksPosition-one click-through rate about 58% lower on keywords with an AI Overview, December 2025 against December 2023Ahrefs, 300,000 keywords, published February 2026Vendor study; an earlier Ahrefs run found 34.5%
AI referrals are smallAI chatbots sent 0.4% of referrals in September 2026, against 91% from search and 8.61% from socialStatcounter, worldwideShare among search, social and chatbot referrals only, not all traffic
AI assistant use is largeChatGPT passed 900 million weekly active usersOpenAI’s own announcement, February 27, 2026Company-reported; later figures from other sources conflict

For your business, this means AI answers already reduce the clicks your best rankings earn, while the visits AI sends in return are still too few to carry a budget on their own.

The two click studies come from Pew Research Center and Ahrefs, and the studies don’t fully agree on the size of either effect. Referral shares move every month, so treat the Statcounter figure as a snapshot of AI search traffic.

That gap makes AEO a visibility and reputation question for now. Being cited shapes what buyers hear about you, even when they never click through, so AEO ROI shows up first in brand demand and only later in AI referral traffic.

So, is AEO worth it for you? Not as a separate budget line if your site still has crawl, indexing or content quality problems, because AI engines draw on that same foundation. It is worth doing now if buyers in your market already ask AI tools for recommendations, or if competitors appear in those answers and you don’t.

A competitor AEO audit can help you compare that visibility before deciding where to invest. The AI search volume keywords tool gives another signal of which questions may be worth investigating, though estimated AI search volumes are not exact demand figures.

A common trap is quoting the biggest number you can find to win the budget. Use dated, sourced ranges, and start by checking your own analytics for AI referral sessions over the last 90 days, so you build the business case for AEO on your own numbers.

// 04 — The Mechanics

How AI Search Works

AI search works by retrieving pages from an index when someone asks a question, then writing an answer from them, a method called retrieval-augmented generation (RAG).

An AI model can know about your content in two ways.

  1. 01Training happens months before anyone asks anything, when a model learns from a large collection of text.
  2. 02Retrieval happens in the moment, when the system searches an index, pulls a few pages and writes an answer from them.

Retrieval matters for AEO, because that’s where citations come from. A link in an answer points to a page the system fetched for that answer, which is why access and indexing come before everything else.

Training still shapes how a model talks about your brand when it answers without searching. Crawler controls let you opt out of future training, but they don’t remove what a model has already learned.

In Google’s case, this retrieval and grounding step runs on the same ranking systems as normal Search, which pull relevant, current pages from the index.

Query fan-out is the second half of the process, where the model runs several related searches at once to gather more results. In Google’s example, a question about fixing a lawn full of weeds fans out into searches about herbicides and weed prevention.

Fan-out explains why a page can be cited for a question it never targeted. A Surfer study of 10,000 keywords and 33,000 fan-out queries found that pages ranking for fan-out queries were 161% more likely to be cited in AI Overviews than pages ranking only for the main query. Depth on a topic counts for more than one page per phrasing.

The chain runs in three steps for every engine, and those steps explain how AI search works from crawl to citation. Crawlers, bots and indexes fetch and store pages, then citation mechanics decide which passages an answer quotes and links. If a page fails at the first step, nothing later can bring it back.

Four engines show how much AI search architectures differ, based on what each vendor states about its own system.

EngineHow it finds pagesCrawler or controlWhat the vendor documents
Google AI Overviews and AI ModeThe Google Search index, with query fan-outGooglebotA page must be indexed, eligible to show a snippet and included in Search generative AI features in Search Console
ChatGPT searchA search crawler, plus visits a user requestsOAI-SearchBot, ChatGPT-UserOAI-SearchBot decides whether sites appear in ChatGPT search answers
PerplexityA search crawler, plus visits a user requestsPerplexityBot, Perplexity-UserPerplexityBot surfaces and links sites and isn’t used for model training
ClaudeA search indexer, plus visits a user requestsClaude-SearchBot, Claude-UserBlocking Claude-SearchBot keeps content out of Claude’s search index

Every row starts with a crawler, which means every engine starts where SEO started in the 1990s, with a bot that has to reach the page before anything else counts.

Engines also cite different sources for the same question. A Writesonic study of 161,286 prompts, collected in May and June 2026, found that no pair of engines shared more than about a quarter of cited domains, with overlap from 12% to 24%. It comes from a single vendor, so treat the exact range with caution, but a citation in one engine clearly says little about the others.

The details of how ChatGPT search works, or how Perplexity ranks sources, change often. Don’t assume that a Google ranking means a ChatGPT citation, or the reverse. Start by searching your three important questions in two engines and noting which sources each one cites.

// 05 — Earning Citations

Getting Cited (What Earns a Place in AI Answers)

An AI answer quotes and links only a handful of pages, and getting cited means making yours one of them. Five areas decide most of it, and each one gets a short overview below.

01

Content structure and format

02

Entity and brand authority

03

Technical access

04

Off-site citation surfaces

05

AI commerce

These five areas work together, but they don’t carry equal weight for every site. Start with the gaps you can verify, rather than assuming one tactic will improve citations across every AI engine.

Content Structure and Format

Content that earns citations says something the other results don’t. Google asks for a unique point of view and warns against pages that restate what is already available elsewhere, and the AI search vendors reward the same thing for a simple reason, since an answer built from five identical pages needs only one of them as a source.

Answering each heading in its first sentence makes a passage easy to quote. Google’s systems don’t need content broken into small chunks, but AI answers quote passages rather than whole pages, so many SEOs still write sections that make sense on their own. Treat that as good editing, with no fixed word count.

Don’t build a separate low-value page for every prompt variation just to manipulate rankings, because that can fall under Google’s scaled content abuse policy. Well-structured content also covers the questions buyers ask, and an answer gap finder can show which ones your pages leave open.

Entity and Brand Authority

An entity is something a search system can identify as a distinct thing, such as your company, product or founder. AI systems build their picture of an entity from every source that mentions it, so a brand described five different ways across the web gives them five different answers to choose from.

Start with consistency. Use the same name, one-sentence description, category and founding facts on your site, social profiles, directories and review platforms. Add Organization schema with links to your official profiles, and keep your About page and author pages specific about who runs the business and who writes the content.

Then claim your knowledge panel and add a Wikidata entry, which feed the structured facts many systems draw on. Pursue Wikipedia only if you meet its notability rules, since a deleted page won’t build entity authority. Check the result every quarter by asking two or three AI tools to describe your company and comparing the answers with your own description.

Technical Access

AI engines can only cite pages their crawlers reach, and each vendor runs separate bots for search, training and user requests. OpenAI says each of its settings is independent of the others, so you can allow search while opting out of training.

These are the crawlers each vendor documents, as checked on October 21, 2026.

CrawlerVendorPurposeIf you block it
OAI-SearchBotOpenAISurfaces sites in ChatGPT searchNot shown in ChatGPT search answers, though navigational links may still appear
GPTBotOpenAICollects content that may train modelsSignals content shouldn’t be used for training
ChatGPT-UserOpenAIVisits pages when a user asksRobots.txt rules may not apply; not used for search inclusion
ClaudeBotAnthropicCollects content that may train modelsFuture content left out of training sets
Claude-UserAnthropicFetches pages when a user asksMay lower visibility in user-directed search
Claude-SearchBotAnthropicIndexes content for search resultsMay lower visibility in Claude search results
PerplexityBotPerplexitySurfaces and links sites in search; not used for trainingMay not appear in Perplexity results
Perplexity-UserPerplexityFetches pages when a user asksUsually ignores robots.txt
GooglebotGoogleCrawls for Search, including AI Overviews and AI ModePages not indexed, so not eligible
Google-ExtendedGoogleToken limiting Gemini training and groundingDoesn’t affect Search inclusion or ranking

Blocking a search crawler removes you from that engine’s answers, while blocking a training crawler leaves search visibility alone, so the decision you make here is mostly about training, not about being found.

A robots.txt file that allows AI search and opts out of model training looks like this:

# robots.txt — allow AI search, opt out of training

User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

Check your CDN and firewall rules too, since they can block bots your robots.txt allows.

Two common add-ons deserve less effort than they get. An llms.txt file won’t help or hurt Google visibility, and an Ahrefs study of 137,210 domains found that 97% of llms.txt files received no requests at all in May 2026.

Schema is less clear-cut. Google says no special markup is needed, while Microsoft’s Fabrice Canel said in March 2025 that schema markup helps Microsoft’s LLMs understand content, so accurate markup is worth keeping, and the structured data validator confirms it matches the visible text.

Then confirm the bots reach you. Server logs show which crawlers visited and which pages they fetched, but a user-agent name is easy to fake, so check the IP address as well.

OpenAI publishes IP lists for each of its bots (searchbot.json, gptbot.json and chatgpt-user.json), Perplexity does the same for PerplexityBot and Perplexity-User, and Google verifies Googlebot through a reverse DNS lookup or its own IP files. Perplexity recommends matching both the user agent and the IP in firewall allow rules, since its ranges change regularly. If a search bot you allowed never appears in your logs, fix that before you touch any content, and recheck your technical AEO setup whenever vendors add new bots.

Off-Site Citation Surfaces

AI answers often quote other platforms, not only brand websites. Semrush tracked more than 230,000 prompts from July to October 2025 and found Wikipedia and Reddit far ahead in ChatGPT, while Ahrefs’ 2026 snapshot of more than 3 million US queries put YouTube first in AI Overviews.

Brand mentions may matter more than links here. An Ahrefs analysis of 75,000 brands found that branded web mentions had the strongest correlation with AI Overview visibility (0.664), against 0.218 for the number of backlinks. Correlation isn’t proof, and the study covered only Google’s AI Overviews, but it points the same way as the citation studies.

The kind of mention still counts. Google warns that chasing inauthentic mentions isn’t worth the effort, and paid placements are easy for both readers and spam systems to spot. Pitch original data, answer questions in communities under your own name and keep review profiles active, so your brand shows up on the off-site sources AI engines read.

AI Commerce

For stores, product data is what AI shopping answers quote. Merchant Center feeds can make products visible in AI responses, so the feed now matters for AI visibility as well as for ads and Shopping results.

Keep prices, stock, shipping and return details identical in the feed, on the product page and in Product schema. AI answers repeat whatever data they find, and a mismatch between your feed and your page can put the wrong price in front of a buyer who never visits your site.

ChatGPT shopping and other assistant shopping features are newer and less documented. Check each vendor’s merchant program and product data requirements before you build anything specific for them, since the rules for product visibility in AI answers are still changing.

Add up the five areas and the score is lopsided. Content structure, entity authority, off-site reputation and product feeds are long-standing SEO work under new names, and technical access is mostly familiar too. The new slice is the crawler matrix, with search, training and user bots from four vendors that you now allow or block one by one.

Across all five areas, the costliest mistake is treating schema or llms.txt as a shortcut. A sounder plan for getting cited starts with access and one strong page.

// 06 — The Engines

The Platforms: ChatGPT, Google, Perplexity and the Rest

AI search platforms are the engines that write answers, and each one decides inclusion through its own index and crawler rules.

The engines also differ in reach. Statcounter’s April 2026 data put ChatGPT at 76.85% of AI chatbot referrals, down from 84.21% a year earlier, with Gemini at 9% and Perplexity at 7.73%.

The table below shows what each vendor documents as the driver of inclusion and the first action to take.

EngineWhat drives inclusionFirst action
Google AI Overviews and AI ModeIndexed, snippet-eligible pages on a site included in Search generative AI featuresCheck indexing, snippet eligibility and the Search generative AI setting in Search Console
ChatGPT searchOAI-SearchBot can reach the pageAllow OAI-SearchBot in robots.txt and at your CDN
PerplexityPerplexityBot can reach the pageAllow PerplexityBot and wait up to 24 hours for changes
ClaudeClaude-SearchBot can reach the pageAllow Claude-SearchBot and Claude-User

Every first action is an access check, the oldest job in technical SEO, so a site that already gets crawling right is eligible on all four platforms. That is also why the 90-day plan below starts there.

→ Google

According to Google’s page on AI features and your website, AI Overviews and AI Mode build on normal Search requirements. Sites must also be included in Search generative AI features in Search Console, a control Google is rolling out gradually.

What people call AI Overview optimization involves a lot of ordinary Search work. For Google AI Overviews and AI Mode, check that your pages are indexed, eligible to show snippets and included under Search Console → Settings → Search generative AI. The setting is on by default, and site owners can opt out without removing their pages from regular Search.

→ ChatGPT

Learning how to rank in ChatGPT starts with access. OpenAI’s crawler documentation says sites that opt out of OAI-SearchBot won’t be shown in ChatGPT search answers. Once access is open, visibility in ChatGPT depends on the same content and reputation work as everywhere else.

→ Perplexity

Perplexity citations depend on PerplexityBot, which Perplexity’s bot page describes as a search crawler that isn’t used for training foundation models. Allowing PerplexityBot is the main setting that decides whether you appear in Perplexity at all.

→ Other engines and agents

Anthropic’s crawler article lists three Claude bots with separate jobs. Google-Extended controls whether Gemini apps may train on or ground answers in your content, and it has no effect on Search.

Agentic browsing, where an AI visits pages for a user, runs through fetchers like ChatGPT-User and Perplexity-User, which may not follow robots.txt. Microsoft Copilot and xAI’s Grok publish less about how they pick sources, so read each vendor’s current documentation before changing anything for them.

Rules for Gemini, Claude, Copilot and Grok change often, and the rise of the agentic web adds a new kind of visitor to plan for. On every AI search platform, big or small, get the basics working before chasing one engine’s quirks, and the FAQ below covers which engine to measure first.

// 07 — Measurement

Measuring AI Visibility Without Fooling Yourself

Measuring AI visibility means tracking whether AI answers mention, cite and describe you correctly, and every method available today has limits.

Here the case for AEO as plain SEO stops holding. Search Console had no report for AI features until June 2026, and even now no standard SEO tool tracks mentions or sentiment inside generated answers, so the skills are familiar while the scorecard is new.

AI answers change from one run to the next, and no single report covers every engine. A sound routine combines four methods, and each one tells you something different and falls short somewhere else.

MethodWhat it tells youMain limit
Prompt set of 20 to 50 prompts, run weeklyWhether you’re mentioned or cited, and how you’re describedAnswers vary between runs, users and days, so results are samples
Search Console Generative AI performance reportImpressions in AI Overviews and AI Mode, by page, country, date and deviceImpressions only, with no clicks, queries or position
Analytics segmented by AI sourceVisits from chatgpt.com, perplexity.ai and other assistantsSmall numbers, and some AI visits arrive without a referrer
Pipeline and brand liftLeads that mention AI tools and growth in branded searchHard to tie to one cause

No single row gives you the full picture, so pick AEO metrics that matter to your business and read the four methods together, never one in isolation.

→ Prompt set

Build it with prompt research that starts from real sales and support questions. Run each prompt in the same engines on the same weekday, log mention, citation and sentiment, and date every run. Repeat each prompt a few times, because one answer is a single sample, and a sound sampling method matters more than the tool you use. An AI query explorer can help you find more questions to include, but keep the final set focused on what your buyers ask.

→ Search Console

Google launched the Generative AI performance report in June 2026 and says it reached all websites worldwide by August 31, 2026. Its data also sits inside the Web search type of the main Performance report, so never add the two together.

→ Analytics

AI referral traffic is small for the majority of today’s sites, so segment it by source instead of watching one total. A spike from a single assistant usually means one answer started citing one page, which is worth checking by hand. In Google Analytics 4, AI referrals land in the Referral channel by default, mixed in with every other site. Create a custom channel group with a rule that matches AI sources such as chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai, and place it above Referral so those visits are counted on their own. OpenAI says ChatGPT adds utm_source=chatgpt.com to referral links, so check your campaign source reports too. Clicks from Google AI Overviews and AI Mode can’t be split out this way, because GA4 counts them as Google organic traffic.

→ Pipeline and brand lift

Add AI tools as an option in your “How did you hear about us?” field, and watch branded impressions in Search Console. Neither gives exact attribution or AEO ROI, but both catch value that never shows up as a referral.

→ Share of voice

Share of voice in AI search is the share of answers in your prompt set that mention your brand rather than a competitor. The figure depends on which prompts you chose, so compare it only with your own past runs. A citation tracker can help you record which answers cite your pages, as long as the prompt set and testing schedule stay the same.

→ Sentiment and accuracy

Log whether each answer describes you positively, neutrally or wrongly, and treat the wrong ones as fixes to make. Fixing a wrong answer starts with its sources. Open the citations under the answer and find where the error came from, which is often an old pricing page, an outdated review or a directory listing nobody updated. Correct your own page first, then ask the third-party site to update its page, and bring your profiles in line so every description matches. Use the feedback option in the engine too, but treat it as a signal rather than a fix. Rerun the same prompt in your next weekly test to see whether the answer changed.

Be wary of any tool that claims special access. Google states that no third-party tool can see its internal ranking or AI systems, and that applies to AEO tools, too.

One week of prompt results is not a trend. Wait for at least four runs, and start this week by writing 20 prompts and running them once to set a baseline.

// 08 — Playbooks

Playbooks by Business Type

AEO playbooks differ by business type, since buyers ask different questions and engines draw on different sources for each one.

Merchant Center feeds and Google Business Profiles are the two business tools Google names as routes into AI responses. Beyond those, priorities depend on how your buyers search.

Each business type has a first priority and a trap that catches it often.

Business typeFirst priorityCommon trap
B2B SaaSComparison, alternatives and use-case pages that answer buyer questionsWriting only about features, so answers cite review sites instead
Ecommerce and DTCAccurate product data in Merchant Center and on product pagesLetting the feed and the page disagree on price or stock
Local and multi-locationA complete Google Business Profile with consistent name, address and phoneThin city pages with only the town name changed
Agencies and consultantsNamed experts with first-hand evidence of their workSelling guaranteed citations to clients
Regulated and YMYL sitesQualified authors and reviewers, plus regular checks of how AI answers describe youIgnoring wrong AI answers about dosage, rates or legal terms

These rows are our editorial starting points, so adjust them to your market. Every first priority in the table, from comparison pages and clean Merchant Center feeds to complete Business Profiles, named experts and qualified authors, was sound SEO advice years before ChatGPT launched, and every trap comes from skipping it.

For B2B SaaS and ecommerce, AI answers to “best tool” or “best product” questions can cite review sites and comparison pages ahead of your own. Keep your own comparison pages fair and current, since an answer can quote an outdated price or feature list as easily as a correct one.

Accuracy carries the most weight in the last row. AEO for healthcare, finance and legal sites means checking AI answers for errors as often as you check them for mentions, since a wrong answer that names you can do more harm than no answer.

Base your AEO strategy on the playbook for your business type that sits closest to your model. Start by listing the five questions buyers ask before they choose you, then check which sources AI answers cite for each.

// 09 — Myth Check

Myths, Hype and Vendor Claims

AEO myths are claims that sound plausible, spread fast and fail when you check them against vendor documentation or dated studies. Each one below starts from the same assumption, that AEO is a brand-new discipline with brand-new rules, because a new discipline is easier to sell than doing the old work well.

MythVerdictEvidence
“AI will kill SEO”FalseAI Overviews run on Search’s ranking systems, and ChatGPT, Perplexity and Claude run search indexes of their own
“You need llms.txt”Not for GoogleGoogle ignores the file, and Ahrefs found that 97% of llms.txt files got no requests at all in May 2026
“Chunk every page for AI”False for GoogleGoogle’s systems need no chunking, though self-contained sections still help readers and quoting
“FAQPage schema earns citations by itself”FalseGoogle needs no special schema, and Bing says schema helps its models understand content, which is not the same as earning a citation
“You can pay or guarantee your way into AI answers”FalseNo vendor documents a paid route into cited sources, and no third-party tool can see inside Google’s AI systems
“One vendor case study proves a method”Weak evidenceA single case study rarely shows a baseline, a sample size or an independent measurer

Read the verdicts together and each myth turns out to sell a new shortcut where the evidence points back to ordinary SEO work, so treat any pitch built on one of them with care.

These myths share a source, which is advice written before the vendors published their own documentation and before independent studies tested it.

AI answers repeat the myths, too. Our October 19, 2026 check of the AI Overview for “answer engine optimization” recommended answers of about 40 to 60 words and schema markup, though no AI vendor requires either one. Short, answer-first paragraphs are good editing, and you can keep them for readers without treating any word count as a rule.

AEO is real as a set of outcomes worth tracking, and its myths fall apart once you read the vendor’s own page, which leaves the old work standing as the part that earns citations.

When you meet a claim that isn’t on this list, test it with three questions.

  1. 01How big was the sample? Count the sites, prompts or queries the claim rests on.
  2. 02What was the baseline? Check what the result was compared with, and over what period.
  3. 03Who measured it? Find out whether the seller of the method or an independent party ran the test.

A claim that can’t answer all three is an opinion, no matter how confident it sounds, and plenty of vendor claims about AEO fail at least one.

// 10 — Operating Model

Teams, Roles and Cadence

The AEO operating model is who owns the work, where it sits in your team and how often it runs.

In a typical company, AEO sits with whoever owns SEO, since the foundation is shared, while content, PR and product marketing supply the rest. AEO marketing works best as part of the existing search plan, with one named owner, which is the “mostly SEO” thesis turned into an org chart.

In a small team, one person can run the whole routine in a few hours a month. Larger teams split it, with SEO owning access and measurement, content owning page updates and PR owning earned coverage.

A monthly cadence in four steps keeps the work manageable.

Week 1

Run the prompt set and pull the Search Console report.

Week 2

Fix the access or accuracy problems the runs exposed.

Week 3

Update one or two priority pages and pitch one piece of earned coverage.

Week 4

Report changes against the baseline and pick next month’s focus.

Repeat it every month, and change the prompt set only once a quarter so results stay comparable.

Who does the work depends on your budget and skills. DIY costs the least cash but the most time, keeps full control with you and starts slowly. A freelancer costs more, shares control and moves faster on defined projects, while an agency costs the most and moves fastest with a full team, with control set by its contract.

If your SEO already runs in-house, DIY is the natural start for AEO, and outside help makes sense once the routine outgrows your team’s hours.

Watch for three red flags in AEO services, which are guaranteed citations, secret methods and claims of access to internal Google metrics. Google warns against tools that make these promises and says no third-party tool has that access.

A clear owner and a steady AEO cadence matter more to your strategy than the size of the budget.

// 11 — Action Plan

Your First 90 Days of AEO

A 90-day AEO plan moves a site from access checks to a measured baseline and better-structured pages in three 30-day blocks.

Start day one with our free AEO readiness check to see where your site stands. The table below then works as an AEO checklist, with an output for every task.

DaysTaskOutputNext guide
1 to 15Decide which search and training bots to allow, check CDN and firewall rules, and confirm Google’s Search generative AI inclusion settingWritten access decision and updated robots.txtHow AI Search Works
16 to 30Write 20 to 50 prompts, run the first weekly test and save a Search Console baselineDated baseline of mentions, citations and impressionsMeasuring AI Visibility
31 to 45Rewrite your top five page intros answer-first and add author biosFive updated pages with named authorsGetting Cited
46 to 60Check that structured data matches visible text and run a page experience checkMarkup and page experience fixes loggedGetting Cited
61 to 75Use one brand description across profiles, plus Google Business Profile and NAP checks for local businessesConsistent entity descriptionGetting Cited
76 to 90Pitch one original data piece or expert quote to relevant publications, then compare runs with the baselineOne pitch sent and a 90-day reportTeams, roles and cadence

The first month is all access and measurement, so by day 30 you know whether engines can reach you and where you stand, before you spend time on content.

Keep the weekly prompt test running through all 90 days, since the final report depends on it.

Don’t expect traffic to move within 90 days. The realistic wins are a clean access setup, a baseline you trust and a handful of pages that answer questions clearly, and every later quarter builds on those.

At day 90, judge the plan by whether each output exists and whether mentions, citations or impressions moved against the baseline. Then use the same 90-day AEO plan format for the next quarter.

// 12 — The Verdict

How Much of AEO Is New

0%

is SEO you already know

That leaves the question this guide opened with. Roughly 70% of the work is SEO you already know, since every engine starts with crawling, four of the five citation areas are familiar work under new names, and the playbooks repeat advice that predates ChatGPT. The new parts are measurement and the crawler choices for engines beyond Google, so that is where new budget belongs.

The engines will keep changing faster than the fundamentals, and so will the labels. The name will change again before the work does, and the surest way to win the newest surface in search is to do its oldest work properly.

// 13 — Common Questions

Frequently Asked Questions (FAQ)

SEO aims to rank pages in search results, while AEO aims to get pages and brands cited inside AI-written answers. Both depend on crawlable pages, useful content and earned trust. Google treats optimizing for its AI features as SEO, and other AI search terms such as GEO describe the same goal from different angles.

GEO stands for generative engine optimization. It describes the same visibility goal as AEO, applied to answers written by generative AI tools such as ChatGPT, Perplexity and Google’s AI features. Google lists both AEO and GEO as terms in use, and day to day, the work behind GEO is the same as AEO under a different name.

Look for tools in five categories, which are prompt tracking, citation tracking, structured data validation, crawler and log checks, and Google Search Console. Search Console and the Schema.org validator are free, and your own server logs cover crawler checks. For prompt and citation tracking, test a few tools on the same prompt set before paying, since results depend on how each one samples answers. Check any vendor’s claims against its documentation, since no tool sees inside Google’s AI systems.

No, AEO doesn’t replace SEO. AI features draw on pages that search engines already crawl and index, so the same foundation earns visibility in both places. AEO adds new outcomes to track, such as citations and brand mentions inside answers, and those outcomes are where the value of AI visibility shows up.

No reliable published benchmark exists yet. Crawl and index delays vary by engine, recrawling can take from several days to several months, and AI answers vary from run to run. Track changes against a dated baseline over at least a quarter, since tracking AI visibility week by week shows noise before it shows trends.

Start from where your audience asks questions. ChatGPT sends the largest share of AI chatbot referrals in Statcounter’s data, though its share fell from 84.21% to 76.85% between April 2025 and April 2026. Pick two engines to measure, often Google and ChatGPT, since where your buyers ask matters more than how AI engines differ.

Yes, on specific questions where they have first-hand knowledge. Content with a unique point of view and first-hand experience beats content that restates what already ranks, and that favors specialists. A small business that answers its customers’ real questions well can be cited next to larger brands, especially with an AEO plan built for its business type.

Not for Google, which ignores llms.txt and needs no special schema for its AI features. An Ahrefs study found that 97% of llms.txt files received no requests at all, so adding one is optional. Bing says schema helps its models understand content, so keep accurate structured data. When either one is sold as a shortcut to citations, it belongs with the other AEO myths.

Start with day one

Check Your AEO Readiness
Free.

See whether AI crawlers can reach your pages and where your site stands before you spend time on content. No signup. No credit card.