How to appear in Google AI Overviews

Google's AI Overviews are not a separate index. They're generated on top of the one you've been optimising for years, which is the best news in this whole field: your existing SEO isn't obsolete, it's a prerequisite.

The system retrieves pages from Google's classic web index, then a Gemini model summarises them and links the ones it used. So the citation pipeline is: rank → get retrieved → have a passage worth quoting → get linked. Miss any step and you're out. Skip the first one and nothing else matters.

Why does classic ranking still decide who gets cited?

Because AI Overviews are grounded in the index. Google isn't asking a model what it knows; it's running a retrieval pass over pages it already ranks and generating a summary from those.

The practical consequence: pages cited in an AI Overview overwhelmingly come from the top of the results for that query or for one of the sub-queries Google fires behind the scenes. If you sit at position 34, the retrieval layer never sees you, and the most beautifully structured passage in the world doesn't help.

This is why "GEO replaces SEO" is wrong for this engine specifically. For ChatGPT, retrieval runs on OpenAI's own index and your Google rank is only loosely correlated. For AI Overviews, your Google rank is the qualifier. Same tactics, same index, new prize. GEO vs SEO goes deeper on where the two actually diverge.

How often do AI Overviews show up, and where?

Depends who you ask, and that spread is itself worth understanding.

Both are correct. They're measuring different keyword universes, and the gap tells you that AI Overviews cluster hard on informational, question-shaped, research-stage queries and appear far less on transactional ones.

Industry matters more than the average. Per BrightEdge, AI Overviews appear on 82% of B2B technology queries, 88% of healthcare, and 83% of education. If you sell software to businesses, this isn't an emerging channel you should keep an eye on. It's most of your funnel's first impression.

What does passage-level optimisation actually mean?

It means Google quotes a chunk, not a page. So structure your content as a set of self-contained, individually-liftable chunks, each one answering exactly one question.

Here's the shape that works:

## How much does an AI visibility tracker cost?

AI visibility trackers range from $29/month to over $800/month. Entry-level
tools like Otterly.AI start at $29/month for 15 prompts, mid-market tools like
Spottlo and Trakkr sit between $39 and $100/month, and enterprise platforms like
Evertune start around $800/month with demo-gated pricing.

| Tool | Entry price | Engines included |
|---|---|---|
| Otterly.AI | $29/mo | 4 (Gemini and AI Mode are paid add-ons) |
| Spottlo | $39/mo | 4, all included |
| Profound | $99/mo | 1 (ChatGPT only at Starter) |
| Scrunch AI | $250/mo | Annual entry |

Three things are happening there and all of them are deliberate.

The H2 is the query. Literally the words someone types. Not "Pricing considerations" — "How much does an AI visibility tracker cost?"

The first sentence answers it completely. With a range, with numbers. That sentence can be lifted, quoted, and shown to a user with zero surrounding context and it still makes sense. That is the entire test.

The table gives structure to lift. Tables are disproportionately cited because they're unambiguous. Every relationship is explicit. A model summarising a table cannot easily hallucinate it.

The passage checklist

Run every section you want cited through this:

Most B2B pages fail four of these seven. Fixing them is a copy edit, not a replatform.

Technical prerequisites

Cheap, boring, and non-negotiable:

• Get indexed. Obvious, and still the most common failure. If it's not in Google's index it cannot be in an AI Overview.
• Server-render your content. Google renders JS, but content behind a client-side fetch is at a real disadvantage in retrieval. Anything you want quoted should be in the initial HTML.
• Don't kill your own snippets. nosnippet and a low max-snippet limit what Google can extract, including for AI Overviews. Setting them is a decision to be less visible.
• Ship the obvious schema. FAQPage, Article, Product, Organization. It doesn't force a citation, but it makes your page trivially parseable. More in schema markup for AI search.
• Keep dates honest and visible. Freshness is a real ranking input for query classes where recency matters, and a stale dateModified on a page you actually updated is a wasted signal.

What does an AI Overview cost you in clicks?

A lot, and you should plan around it rather than pretend otherwise.

Pew Research found that only 8% of users click any search result when an AI summary is shown, versus 15% when it isn't. Just 1% click a link inside the summary itself. And Ahrefs, in a 300,000-keyword Search Console study, measured the #1 organic result losing 58% of its CTR when an AI Overview is present.

Read those two together and the picture is unambiguous. The AI Overview absorbs the click. Being cited inside it gets you a 1% click-through. Position 1 underneath it gets you a heavily-taxed version of what you used to get.

So why bother?

Because the mention is now the asset. Someone who reads "Spottlo, Otterly and Trakkr are the affordable options" in an AI Overview and doesn't click has still learned your name, and they'll type it directly later or recognise it in a shortlist. That's brand, delivered at the exact moment of consideration, and it doesn't show up in your analytics at all.

And when they do click, they're better. AI-referred visitors convert 54% better than non-AI traffic on US retail sites, where AI referrals grew 138% YoY. Fewer, warmer. The right response is to stop measuring this channel purely in sessions, which is what zero-click search strategy is about.

How do I track AI Overview citations?

Search Console doesn't separate them out. Impressions from AI Overviews are folded into your regular numbers, and there's no "cited in AI Overview" dimension. So you have two options.

Manual: take your 20-30 most valuable queries, run them in an incognito window, and record whether an AI Overview appeared and whose domains it cited. Repeat monthly. It's tedious and it works.

Automated: run a fixed prompt set on a schedule and log the citations. That's what Spottlo does — AI Overviews alongside ChatGPT, Perplexity and Gemini, weekly, with mention rate, position and share of voice against the competitors you name. All four engines are on every plan, which is not true of most tools in this category (see the comparison).

Either way, the metric that matters is: for the questions my buyers ask, how often is my domain one of the cited sources, and who's there instead of me?

What to do next

1. Pull your top 30 commercial queries and check which trigger an AI Overview. If you're in B2B tech, expect most of them to. That's your target list.
2. For each triggering query, check who Google cites. If it's a listicle you're absent from, that's a content gap. If it's a competitor's page, read it and note which passage got lifted.
3. Rewrite your ranking pages passage-first. Query as the H2, answer in the first two sentences, a table where the data is comparative. Use the checklist above.
4. Verify you haven't suppressed your own snippets with nosnippet or a restrictive max-snippet.
5. Set a baseline you can re-measure. Run a free AI visibility report and see where you currently stand across AI Overviews and the chat engines before you change anything.