Ahrefs measured the same thing twice, eight months apart, and got 76% and 38%. The question was how many pages cited in a Google AI Overview also rank in the top 10 for the same search. In July 2025 the answer was about three in four. By March 2026 it was closer to two in five.
Both numbers are correct. Together they show that ranking and being cited are drifting apart, which matters to anyone paying for "search visibility." The rules that decide who gets mentioned are moving faster than the vocabulary used to sell them. SEO, AEO and GEO are three labels for three related jobs, and a company that does not run a search team still has to decide which ones deserve money.
What do SEO, AEO and GEO each mean?
SEO is search engine optimization, the work of earning a high position in the list of links a search engine returns. AEO is answer engine optimization, the work of being the source a search engine or assistant lifts a direct answer from. GEO is generative engine optimization, the work of being cited or recommended inside a written response that an AI model composes.
The three overlap heavily, and practitioners disagree about where one ends. The term GEO comes from a 2023 paper by researchers including Pranjal Aggarwal and Karthik Narasimhan, accepted at KDD 2024. They reported that rewriting content in specific ways could raise its visibility in generative engine responses by up to 40%, measured on their own benchmark of queries rather than on live business results.
One warning on vocabulary: some writers use AEO to mean agent engine optimization, which is about software that buys on a person's behalf. That is a separate job, and we cover it in our piece on agent engine optimization. Here AEO means answer engines.
| SEO | AEO | GEO | |
|---|---|---|---|
| What it aims for | A high position in the list of links a search engine returns | Being the source a search engine or assistant lifts a direct answer from | Being cited or recommended inside a written response an AI model composes |
| Where it shows up | The list of links | A featured snippet or a voice reply | A response from ChatGPT or a Google AI Overview |
| How you measure it | Rank and clicks | Whether you are the source of the direct answer | Mentions and citations |
| What a B2B company should do | Keep the fundamentals: pages that can be crawled, load fast and answer the question | Put the direct answer in the first two sentences under each heading | Be named where assistants read, such as comparison pages, review sites and your own documentation, and track mentions next to sessions |
Where do your buyers get answers now?
Increasingly from a written summary, not from a page they visit. The best controlled measurement of what that does to clicks is from Pew Research Center. In a March 2025 browsing study of 900 US adults and 68,879 Google searches, people clicked a traditional result link in 8% of visits when an AI summary appeared. When it did not, they clicked in 15%.
Only 1% of visits included a click on a link inside the summary itself. People also ended their browsing session more often after a summary, 26% of visits against 16%.
For B2B the picture has a second layer. Forrester's 2026 State of Business Buying, based on nearly 18,000 global buyers surveyed in 2025, found 94% of business buyers use AI during the buying process.
Gartner counted differently: its survey of 645 B2B buyers, fielded in August and September 2025, found 45% had used generative AI, and 69% prefer to validate what it tells them with sales reps. The gap comes from what each asked: any AI use in the purchase versus generative AI specifically. Either way, an assistant may draft the shortlist, but people still check it.
Do the pages that rank also get cited?
Less often than they used to, and the honest answer is that the two studies count different things. The July 2025 Ahrefs analysis covered 1.9 million citations from 1 million AI Overviews and found 76.1% of cited pages ranked in the top 10. The March 2026 analysis covered 863,000 keyword results and 4 million cited URLs and found 37.9% did.
The 2025 study counted only the top three citations per answer. Ahrefs also attributes the drop to a newer Gemini model and to query fan-out, where one question is split into several related searches and sources come from all of them. In the newer data, 31.2% of citations came from pages ranked 11 to 100, and 31.0% from pages beyond the top 100.
Read together, the studies say ranking is still the largest single source of citations, and it is no longer most of them. A page that ranks gives you a good chance of being cited. A page that does not rank is not shut out.
What should a B2B company change, and what should it keep doing?
Keep the SEO fundamentals, because every other layer sits on them. Pages that cannot be crawled, load slowly or answer nothing will not be found by a search engine or an assistant. Our review of what the best SEO optimization companies do differently covers that base, and nothing here replaces it.
Then change four things, in this order:
- Answer first on the pages that matter. Put the direct answer to the question in the first two sentences under each heading, then explain. This helps a searcher, a snippet and an assistant alike, which is why AEO and GEO advice converges here.
- Be named where assistants read. Comparison pages, review sites, your own documentation and partner pages all feed answers. This is our inference from the citation spread above, not a measured rule.
- Report differently. If clicks per search fall by roughly half when a summary appears, traffic alone will understate how often you are seen. Track mentions and assisted pipeline next to sessions.
- Look at the tools, but buy with care. Our list of generative engine optimization tools covers what they measure. No tool or agency can guarantee a mention, because the models decide.
What to stop: paying a premium for "GEO" work that is SEO under a new name, and rewriting everything at once. Pick your ten pages closest to revenue and start there.
What is worth automating here?
The recurring, boring part: finding out where you stand. A weekly check takes a fixed set of 20 or so questions your buyers actually ask, runs them through the main assistants, and records whether your company is named, who else is, and which pages were cited. Done by hand it is tedious enough that it happens twice and then stops.
This is the kind of routine we build and run for teams as marketing workflow automation: you describe it in plain language, and we build it, keep it running and adjust it when the assistants change. Most of it is tested code that asks the questions and stores the answers. AI is used for one narrow step, reading each response and deciding whether your company was mentioned.
The cost of that model is the usual one. You hold a weekly report, not a codebase, and the check stops when the plan does. A company that wants to own the tooling should buy it from a development shop instead.
What can nobody measure yet?
Whether being cited by an assistant moves revenue. No study we found links AI mentions to closed deals for B2B companies, and answers vary from one run to the next, so a single check proves little. The Aggarwal paper tested rewrites on a benchmark, not on a sales pipeline.
That leaves a decision rule you can apply without any vendor. If a change makes a page clearer for a buyer, do it, because it helps in every channel. If it only makes sense for a model and would confuse a person, wait until someone shows you the result in your own numbers.
Frequently asked questions
What is the difference between GEO and SEO?
SEO aims to rank a page in the list of links a search engine returns. GEO aims to get your company cited or recommended inside a written answer that an AI model composes, such as ChatGPT or a Google AI Overview. They share fundamentals like clear, crawlable content, but they are measured differently: rank and clicks for SEO, mentions and citations for GEO.
What is the difference between AEO and GEO?
AEO (answer engine optimization) focuses on being the source of a direct answer, such as a featured snippet or a voice reply. GEO (generative engine optimization) focuses on being cited inside a longer response an AI model writes. The overlap is large, and many practitioners use the two terms interchangeably.
Does GEO replace SEO?
No. Ahrefs found 37.9% of AI Overview citations in March 2026 also ranked in the top 10, so ranking is still the largest single source of citations. SEO fundamentals such as crawlable pages and clear answers also support GEO, so dropping SEO usually weakens both.
What is GEO in marketing?
GEO, or generative engine optimization, is the practice of making content more likely to be cited by AI search tools and assistants. The term comes from a 2023 research paper by Aggarwal and colleagues, who reported visibility gains of up to 40% on their test benchmark.
How do I know if AI assistants mention my company?
Ask the main assistants the questions your buyers ask, and repeat each several times because answers vary. Record whether you are named, who else is, and which sources were cited. A weekly routine like that can be automated, and the trend over several weeks tells you more than any single answer.