01Notes · 29 August 2026
Why Your Brand Is Missing From Google AI Mode
Quick answer
The first thing to check is whether you are measuring the right surface. AI Mode and AI Overviews are different products, and a reading from one tells you almost nothing about the other.
Google reported AI Overviews at roughly 2.5 billion monthly active users and AI Mode at around 1 billion at I/O in May 2026. Both are large enough to warrant separate measurement.
Key takeaways
- Sample AI Mode separately. Folding it into an AI Overviews reading is the most common error on Google’s surfaces.
- It handles longer, conversational queries and runs its own follow-up reasoning, so a page that satisfies a short query may not satisfy a multi-part one.
- It is the most geo-localised engine measured. A 2026 study scored its cross-market consistency at 0.146.
- Ranking helps but does not decide. 55.2% of AI citations came from outside the traditional top ten.
- Answer in the first 100 words. 22.1% citation rate against 8.7% otherwise.
- Nothing AI-specific is required. No llms.txt, no chunking, no special schema — Google names all three as unnecessary.
Why it is a different surface
AI Overviews sit above search results and summarise. AI Mode is a conversational experience that handles a query, reasons about it, and often runs follow-up work of its own before answering.
Those are different jobs, and they select sources differently. A page that provides a clean definitional answer will do well in an Overview. The same page may be passed over in AI Mode if the query has three parts and the page only answers one.
| AI Overviews | AI Mode | |
|---|---|---|
| Scale | ~2.5bn monthly users | ~1bn monthly users |
| Query type | Shorter, often informational | Longer, conversational, multi-part |
| Where it appears | Above results | Its own experience |
| Cross-market consistency | Higher | 0.146: the most localised measured |
| How to sample | Trigger rate and citation, separately | Separately again, locale fixed |
| Common error | Assuming it always triggers | Assuming an Overview reading covers it |
The multi-part query problem
This is the part worth acting on. AI Mode gets asked questions like "which of these tools works with our existing stack, and what does it cost for a team of twenty"two or three questions in one sentence.
A page answering only the first part is a partial source. The system will find a page that answers more of it, or assemble the answer from several. That is the mechanism behind a finding we cite often: pages serving eight or more related queries reached the AI top ten 43.3% of the time against 15.5% for single-query pages.
The practical consequence is consolidation, not publication. Three thin pages each answering one part of a question lose to one page answering all three. Most sites already have those three pages.
Why a US reading tells you nothing about the UK
A June 2026 cross-market study measured how much citation sets overlapped between markets and gave AI Mode a consistency score of 0.146: the most geo-localised of the engines tested.
If you operate in more than one market and you sampled AI Mode in one of them, you have a reading for that market only. The market-level detail is here.
The things that actually help
- Answer in the first 100 words. The strongest structural signal measured: 22.1% citation rate against 8.7%.
- Cover the query family. One page that answers the whole question, not three that each answer a third.
- Be reachable. Googlebot access, renderable content, no JavaScript-only critical copy.
- Be specific. Named entities, dated figures, stated sources.
What does not help: llms.txt, content chunking, AI-specific schema. Google names all three as unnecessary, and we went through that guidance in detail here.
What this means for you
- Check you are sampling AI Mode at all. Many reports labelled "AI visibility" only cover Overviews.
- Consolidate before you publish. The multi-part query rewards one complete page over three partial ones.
- Fix the locale in your sampling and record it, because this is the surface where it matters most.
02Common questions
Questions this raises
Is AI Mode the same as AI Overviews?
No, and treating them as one is the most common measurement error on Google's surfaces. They are different products with different behaviour. Google reported AI Overviews at roughly 2.5 billion monthly active users and AI Mode at around 1 billion at I/O in May 2026.
Why would I appear in one and not the other?
Because AI Mode handles longer, more conversational queries and runs its own follow-up reasoning. A page that satisfies a short informational query may not satisfy a multi-part one.
Does ranking help in AI Mode?
It helps, as it does across Google's surfaces, but less than people expect. In 2026 research 55.2% of AI citations came from outside the traditional top ten.
Is AI Mode the most localised surface?
In the data we have, yes. A June 2026 cross-market study gave it a consistency score of 0.146: the most geo-localised of the engines measured. A result in one market predicts very little about another.
How do I sample AI Mode properly?
Separately from AI Overviews, in a fresh session, with the locale fixed and recorded. Blending the two produces a number that describes neither.
What actually helps?
The same things that help everywhere: answering in the first 100 words, covering a query family rather than one keyword, factual specificity, and being reachable by Googlebot.
Sources
Every figure on this page comes from one of the following. Where a claim is ours rather than a source’s, it carries an evidence label instead.
- Google Search Central — Optimizing your website for generative AI features (15 May 2026, updated 10 July 2026), and Google I/O announcements, 19 May 2026. Primary
- Rankability — State of AI Search, 1,645 citation observations across 31 topics, 2026. Source for the 24.1% cross-platform overlap, the 73.5% ChatGPT exclusive citation rate and the first-100-words finding. Independent study
- Searlixa sampling across client engagements, labelled Observed with sample and date. Our own
Ask an assistant
Hand this page to an assistant and ask it to summarise or challenge the argument. If it gets something wrong about us, that is itself a useful finding.
→Next step
Find out which Google surfaces you are on.
AI Overviews in detail · Diagnose Overviews absence · More notes