AEO fundamentals
AI search optimization: one term, five surfaces, two mechanisms.
“AI search” is not one place. It is Google's AI layer, ChatGPT with search, Perplexity, Copilot and Gemini — each reaching your site a different way, and all of them answering either from live pages or from what the model already knows. This page is the map: what each surface is, how it reads the web, and which to work on first.
Updated 2026-09-19
The phrase “AI search” makes it sound like one place, and it is at least five. A question typed into Google may get an AI Overview above the links. The same question asked in ChatGPT may trigger a live search and come back with citations. Perplexity cites by default. Copilot sits on Bing. Gemini sits on Google and on the phone. AI search optimization is the work of being in all of those answers, and it is more tractable than it sounds, because underneath five surfaces there are only two ways an answer gets made. This page is the map; the per-engine playbooks linked below are the territory.
Two mechanisms under five surfaces
Every AI answer about your business is produced one of two ways. Either the system retrieves live pages from a web index and composes an answer from them — in which case the question is whether your page was findable, readable, and the best answer to what was asked — or it answers from what the model already knows, which was shaped by everything written about you over a long time. Most surfaces blend both. The framing is ours, and it matters because the two paths move on completely different clocks: the retrieval path in days and weeks, the model path in months.
Optimization means working the retrieval path first, because it is the one you can move, while the slower path accumulates from the same work. Nothing on this page is a claim about any engine's internals beyond the documented fact that these products search the live web.
The five surfaces
| Compared on | Surface | Where the answer appears | How it reaches your site | Who should work on it first |
|---|---|---|---|---|
| Google AI Overviews and AI Mode | Google AI Overviews and AI Mode | Above the organic results, on the searches that trigger one. | Google's own index. If you are indexed and rank for the underlying query, you are eligible; no special markup is required. | Anyone whose customers still start on Google — which is nearly everyone. The classic SEO layer is the entry ticket. |
| ChatGPT with search | ChatGPT with search | Inside the conversation, when the model decides to look something up or the user asks it to. | A live web index reached through OAI-SearchBot. Independent testing found roughly 87% of its citations match Bing's top results — a measured correlation, not a documented mechanism. | Businesses whose buyers ask conversational, multi-part questions: comparisons, recommendations, 'who should I use for…'. |
| Perplexity | Perplexity | Every answer, with citations shown by default. | Its own live retrieval over the web; the most citation-forward surface, so quotable pages matter most here. | Research-heavy purchases and anything where the buyer wants to see sources. |
| Microsoft Copilot | Microsoft Copilot | Bing's answer layer and the Copilot apps. | Bing's index directly. Bing standing is readable and moves in days, which makes it the earliest indicator on this list. | Anyone — because Bing standing is the fastest signal you can watch, whatever surface you ultimately care about. |
| Gemini | Gemini | Google's assistant, in the app and increasingly inside Google products. | Google's search grounding plus the model's own knowledge; behaves like a conversational cousin of AI Overviews. | Consumer-facing local businesses, where Android and Google Maps are the front door. |
Google AI Overviews and AI Mode
Surface
Google AI Overviews and AI Mode
Where the answer appears
Above the organic results, on the searches that trigger one.
How it reaches your site
Google's own index. If you are indexed and rank for the underlying query, you are eligible; no special markup is required.
Who should work on it first
Anyone whose customers still start on Google — which is nearly everyone. The classic SEO layer is the entry ticket.
ChatGPT with search
Surface
ChatGPT with search
Where the answer appears
Inside the conversation, when the model decides to look something up or the user asks it to.
How it reaches your site
A live web index reached through OAI-SearchBot. Independent testing found roughly 87% of its citations match Bing's top results — a measured correlation, not a documented mechanism.
Who should work on it first
Businesses whose buyers ask conversational, multi-part questions: comparisons, recommendations, 'who should I use for…'.
Perplexity
Surface
Perplexity
Where the answer appears
Every answer, with citations shown by default.
How it reaches your site
Its own live retrieval over the web; the most citation-forward surface, so quotable pages matter most here.
Who should work on it first
Research-heavy purchases and anything where the buyer wants to see sources.
Microsoft Copilot
Surface
Microsoft Copilot
Where the answer appears
Bing's answer layer and the Copilot apps.
How it reaches your site
Bing's index directly. Bing standing is readable and moves in days, which makes it the earliest indicator on this list.
Who should work on it first
Anyone — because Bing standing is the fastest signal you can watch, whatever surface you ultimately care about.
Gemini
Surface
Gemini
Where the answer appears
Google's assistant, in the app and increasingly inside Google products.
How it reaches your site
Google's search grounding plus the model's own knowledge; behaves like a conversational cousin of AI Overviews.
Who should work on it first
Consumer-facing local businesses, where Android and Google Maps are the front door.
The Bing correlation is Seer Interactive's 2026 measurement of where ChatGPT's cited pages also rank — a third-party correlation, not an OpenAI-documented mechanism. The Google statement on markup is Google's own. The 'who first' column is our judgment.
How we read each one, stated the way the product states it
- ChatGPT — scanned through the live consumer interface, daily
- Gemini — scanned through the live consumer interface, daily
- Perplexity — answers captured daily through its API
- Google AI Overviews — read from live Google results
- Microsoft Copilot — read from Bing's answer layer
- Claude — model mode: how the model understands your brand long-term (labeled, never counted as a live capture)
The last entry is the honest edge of the map: a model asked what it believes about a brand is the second mechanism, not a live answer, and it is labeled that way rather than counted as one.
Why Google's layer comes first for almost everyone
48%
of Google searches now return an AI Overview above the organic results
Source: Similarweb, March 2026
61%
fewer organic clicks on searches where an AI Overview answers first
Source: Similarweb, 2026
Which is the argument in one line: the surface sits on the search your customers already use, and when it answers first, the links under it lose most of their clicks. Being in the answer is the only defense against being under it.
It is also the surface where the least new work is required, because it inherits your classic ranking: if you are indexed and rank for the underlying query, you are eligible, and Google states no special markup is required. The work is the same readable, question-shaped page that serves everything else.
Why Bing gets watched throughout, whatever you care about
Not because your buyers are on Bing. Because Bing's standing for a page is readable in days rather than weeks, and because independent testing found roughly 87% of ChatGPT's citations match Bing's top results (Seer Interactive, 2026). That is a measured correlation and not a mechanism OpenAI documents, and it is still the earliest indicator available: a page submitted to Bing via IndexNow and checked there tells you something useful long before any assistant cites it. Bing is never the goal; it is the gauge.
The crawler you may have blocked without knowing
ChatGPT with search reaches the web through OAI-SearchBot. GPTBot is primarily training-data collection, and blocking it does not remove a site from ChatGPT Search — but a lot of sites got a blanket block added years ago by a web company that never distinguished the two jobs. Anthropic splits the same way, with Claude-SearchBot for search appearance and ClaudeBot for training. Whatever you decide, decide it knowing which surface each rule affects.
The per-engine playbooks
- Google AI Overviews optimization
- ChatGPT: how to get recommended
- Perplexity: the citation-first surface
- Bing and Copilot: the early gauge
- AEO vs GEO — the vocabulary, if you need it; and what AEO is, the plain definition.
08 · FAQ
The questions people ask about the surfaces.
What is AI search optimization?
Is it one thing or five things?
Which surface should I work on first?
Does the same work serve all five?
Do I need special schema for AI search?
How fast does any of this move?
Keep reading
Engine playbooks
Google AI Overviews
The data-dense guide to Google's three AI surfaces — Overviews, AI Mode, and Gemini — in one map.
Engine playbooks
ChatGPT SEO
The practitioner playbook for the biggest engine — built around the Bing backbone nobody else leads with.
Engine playbooks
Perplexity SEO
The one engine where every answer carries citations — its source patterns and how to enter them.
AEO fundamentals
What is AEO
The plain-English definition, the mechanics, and why the acronym war (AEO vs GEO vs AI SEO) doesn't matter.
See what each surface says about you today.
The free audit asks ChatGPT, Gemini, and Perplexity ten buyer questions about your business — 30 live answers in about two minutes, three of them shown word for word with every name highlighted. Just an email, no account, no card.
Free audit first: see exactly where you stand before paying anything. 7-day trial, no card.