AEO and GEO: The Agency Guide to Getting Cited by ChatGPT, Perplexity and AI Overviews

Search used to end with a list of ten links. More and more often it ends with an answer. Google writes a summary at the top of the page, ChatGPT and Perplexity write a paragraph with a few citations, and the person reading it may never visit a website at all. For an agency, that changes what "winning" means for a client.

This guide is the hub for everything we publish on the topic. It explains what answer engine optimization (AEO) and generative engine optimization (GEO) are, what the evidence says actually works, and how to turn it into a repeatable service. Each section links to a deeper article.

What are AEO and GEO?

AEO and GEO are two names for the same goal: becoming the source an AI system uses and cites when it answers a question. AEO grew out of optimising for featured snippets and now covers Google's AI Overviews and AI Mode. GEO is the term researchers and vendors adopted for generative engines such as ChatGPT, Perplexity, Gemini and Claude.

The distinction matters less than the shared idea. In traditional SEO, success is a ranking and a click. In AEO and GEO, success is also a mention of the brand inside the answer and a cited link to the client's page, whether or not the person clicks.

Why does this matter for agencies now?

Because the clicks that agencies have always reported are shrinking on the queries where answers appear. Studies published in 2025 and 2026 put the drop in organic click-through rate under an AI Overview somewhere between about a third and more than half, and Pew found people clicked a result roughly half as often when an AI summary was shown. We compare every major study in how much traffic AI Overviews really take.

That creates two problems for agencies. Traffic reports start to look worse even when the work is good, and clients start asking whether SEO still matters. AEO gives you a way to keep winning on those queries and a way to show it.

How do AI search engines choose what to cite?

They retrieve candidate pages, read them, and cite the ones that best support the answer. The details differ by engine. Perplexity runs a live web search for nearly every question and cites sources inline. ChatGPT combines what its model already knows with live search when it decides a search is needed. Google's AI features draw on Google's own index and ranking systems.

Three consequences follow for every engine:

• A page has to be reachable. If an AI crawler is blocked or cannot read the page, it cannot be cited.
• The relevant passage has to stand on its own. Engines lift a sentence or paragraph, so the answer must be clear without the surrounding page.
• Trust signals still matter. Engines favour sources that look authoritative and consistent with the rest of the web.

We go deeper, with a test you can run yourself, in how ChatGPT and Perplexity choose sources. Google's conversational mode has its own quirks, covered in Google AI Mode SEO.

The AEO playbook: what actually works

The honest answer is that the fundamentals do most of the work, and some popular tactics have weak evidence behind them. Here is the playbook we use, in priority order.

1. Make sure AI crawlers can reach and read the site

AI systems use several crawlers with different jobs: some collect training data, some build search indexes, and some fetch a page when a user asks. Blocking the wrong one can remove a client from AI search entirely. Our reference on AI crawler user agents lists them and shows the robots.txt rules.

Rendering is the other half. A study by Vercel and MERJ found that the major AI crawlers, apart from Google's and Apple's, do not execute JavaScript. If a client's content only appears after scripts run, most AI crawlers see an empty page. See can AI crawlers read JavaScript?.

2. Write answer-first content

Put the answer in the first one or two sentences of a section, then support it. Use clear headings phrased the way people ask questions, tables for comparisons, and short definitional sentences that can be quoted on their own. This is ordinary good writing made stricter. Our guide to answer-first content covers the patterns with examples.

3. Build the entity, not just the page

AI systems form a view of a brand from everything written about it: the client's site, review platforms, directories, news coverage and community discussion. Consistent facts, clear "about" information and mentions on trusted third-party sites help an engine understand who the client is and when to recommend them. See entity SEO for LLMs.

4. Treat schema and llms.txt as hygiene, not boosters

This is where the evidence is most often overstated. A large Ahrefs study in May 2026 found that adding JSON-LD schema did not measurably increase AI citations for pages that were already being cited. Schema still helps with rich results and clarity, so keep it, but do not sell it as a citation lever. We review the research in does schema markup get you cited by AI?.

The same applies to llms.txt. Adoption is growing, but analyses of server logs find that AI crawlers rarely request the file, and there is no evidence that having one improves citations. It costs little to publish, so it is fine as hygiene. The data is in does llms.txt actually work?.

5. Earn the citation with something only the client has

Original data, first-hand testing, clear pricing, expert commentary and specific local knowledge are hard for an engine to find elsewhere. Pages that offer them are more likely to be the one source worth citing.

How do you measure AEO?

You measure it with three layers of data, because no single tool sees everything.

For the visibility layer, see AI share of voice. For traffic, see how to track ChatGPT and Perplexity traffic in GA4. For turning all of it into a client conversation, see proving SEO value when clicks fall.

How to package AEO as an agency service

Clients understand AEO quickly once they see an AI answer that recommends a competitor. The work fits naturally into an existing SEO retainer in three phases.

1. Baseline (weeks 1–2). Build a prompt set for the client's category, record where they and their competitors are mentioned and cited today, and fix any crawler access problems.
2. Improve (ongoing). Rewrite priority pages answer-first, fill content gaps where competitors are cited, strengthen entity signals and publish original data.
3. Report (monthly). Show citation share against competitors, AI referral traffic and Search Console AI impressions next to the usual SEO metrics.

Pricing this work is its own topic; see SEO retainer pricing in 2026.

Common mistakes to avoid

• Chasing a single tactic. No file, tag or markup type unlocks AI citations on its own.
• Blocking search crawlers by accident. Blanket AI-bot blocks in robots.txt often remove search access along with training access.
• Reporting only clicks. A client cited in every relevant answer can still see falling sessions. Report both.
• Ignoring the rest of the web. An engine that never sees the brand mentioned elsewhere has little reason to trust it.

A 90-day AEO roadmap for a new client

A clear plan helps clients understand what happens when, and it gives you checkpoints to report against.

Days 1–15: baseline and access. Build a prompt set of 30 to 60 questions across the buying journey. Record mentions and citations for the client and three competitors on each engine. Audit robots.txt for AI search and user agents, and check that priority pages render their content in the raw HTML. Fix anything blocking access first, because it caps every other result.

Days 16–45: priority pages. Pick the ten pages most relevant to the prompts where competitors are cited and the client is not. Rewrite them answer-first, add the specifics people ask about, update facts and dates, and make sure each page owns one clear topic. Fix the canonical facts on the about page and the main owned profiles.

Days 46–75: gaps and corroboration. Publish content for the prompts where the client has no relevant page at all. Start building presence on the third-party sites that engines cite for the category: review platforms, directories, comparison articles and trade publications.

Days 76–90: first report and next cycle. Re-run the full prompt set, compare against the baseline, and report citation share alongside AI referral traffic and Search Console AI impressions. Use the gaps that remain to plan the next quarter.

Expect the fastest movement on Perplexity and ChatGPT search, which retrieve live, and slower movement in unprompted brand recommendations, which depend on the brand's wider reputation.

How Rankwyn helps

Rankwyn's AEO Citation Radar tracks mentions and citations for each client across ChatGPT Search, Perplexity and Google's AI features, and its audits flag crawler and rendering problems across the whole portfolio. Fixes can be staged for approval and applied to the site from one place.

The bottom line

AEO and GEO are SEO with a new finish line. Keep the fundamentals strong, write so a single passage can answer the question, build the brand's presence across the web, and measure citations as well as clicks. Use the linked guides in this hub to go deeper on each step.