About this digest
This digest compiles published sources on how AI assistants and Google's AI features pick the pages they cite. It contains no data from Zunkiree Labs. Where a number comes from a company that sells search or listing tools, we say so.
Our own readings are labelled as ours. Treat every percentage as a prompt to check your own position, not as a benchmark to hit.
What Google documents
Google's page on AI features and your website is direct. It says there are no additional requirements to appear in AI Overviews or AI Mode, and no other special optimizations are necessary. A page must be indexed and eligible to appear in Search with a snippet, and meeting the requirements does not guarantee crawling, indexing or serving.
- No new files or markup. Google says you do not need new machine-readable files, AI text files, extra markup or special structured data for these features.
- The basics still apply. Allow crawling, link internally, give a good page experience, keep important content as text, and make structured data match the visible text.
- Measurement. AI Overviews and AI Mode traffic is counted in the normal Web search type in Search Console and is not reported separately.
- Control. Googlebot's robots.txt rules govern access, and nosnippet, data-nosnippet, max-snippet or noindex limit what is displayed.
What the largest published citation sample found
A Yext Research study, "AI Citation Behavior Across Models (Q4 2025)", reported by Lauryn Chamberlain on September 22, 2026, analysed 17.2 million distinct AI citations across four models and seven sectors. It grouped the sources into four buckets.
| Source type | Share of citations |
|---|---|
| Brand-owned websites | 46.89% |
| Managed third-party listings | 43.71% |
| Reviews and social | 5.48% |
| News, forums and other | 3.92% |
Brand sites and managed listings together supplied about 90.6% of citations. Listings made up 54.53% of the distinct cited URLs, while brand websites averaged 4.31 citations per URL against 2.46 for listings. The study comes from a company that sells listing management, so the exact figures are directional.
Different assistants, different sources
The split changes sharply by model in the same study:
| Model | Brand-owned site | Managed listings |
|---|---|---|
| Gemini | 51.62% | 39.15% |
| Perplexity | 46.11% | 43.41% |
| Claude | 41.27% | 45.93% |
| ChatGPT | 39.55% | 53.36% |
Other analyses point the same way. As reported in the secondary sources Onely and Superlines, Ahrefs found that only about 12% of URLs cited by AI tools overlap with Google's top-10 organic results, and one audit reported about 11% overlap between the domains ChatGPT and Perplexity cite. We have not checked those underlying datasets ourselves.
How far to trust the numbers
- Vendor-produced. The Yext study comes from a listing-management company, and most AI-visibility statistics in circulation come from GEO and tracking tool vendors.
- Consistent in direction. Studies with different samples and methods keep reaching the same qualitative conclusion: assistants cite a different mix of sources than Google ranks, and structured profile pages carry real weight.
- Not established. Any specific percentage. Sample windows differ, "citation" is defined differently across studies, and none of these datasets has been independently replicated.
What this means for visibility
Our own suggestion, not a finding from the studies:
- Claim and correct your profiles on the directories and review sites your buyers use, and keep the description identical to your website's wording.
- State plainly what you do in one self-contained sentence near the top of key pages, because assistants lift passages that make sense out of context.
- Add Organization structured data with sameAs that lists every profile you control, so your site and listings are linked.
- Check what assistants say today. Ask the questions your buyers ask and note which sources are cited. That audit is more useful than any published average.
- Do not buy a special route into Google AI Overviews. Google says none is required.
For the mechanics, see our answer engine optimization whitepaper and our step-by-step guide to getting cited by AI engines.
Limits of this digest
- It summarises published sources and was not independently replicated.
- The headline figures come from one vendor study and secondary reports of other studies.
- Assistants change quickly, so figures from late 2025 data may not describe today's behaviour.
- Google's documentation describes its own features only, not other assistants.
Sources
- Yext, Lauryn Chamberlain, Cracking the ChatGPT Code: How to Grow Brand Visibility in OpenAI, September 22, 2026, reporting Yext Research, AI Citation Behavior Across Models (Q4 2025)
- Google Search Central, AI features and your website
- Onely, What Influences Brand Visibility in AI Search? A Practical Guide for 2026 (secondary)
- Superlines, AI Search Statistics 2026: 60+ Data Points on Visibility, Citations, and Traffic (secondary)
Frequently asked questions
Do I need special markup to appear in Google AI Overviews?
Google's documentation says there are no additional requirements to appear in AI Overviews or AI Mode, and no other special optimizations are necessary. A page must be indexed and eligible to show a snippet.
Where do AI answers get their sources?
In one large vendor-published sample of 17.2 million citations, brand-owned websites supplied 46.89% and managed third-party listings 43.71%. The split varied by model, and the figures are directional, not audited.
Does ranking in Google mean being cited by AI assistants?
Not necessarily. Ahrefs has reported that only about 12% of URLs cited by AI tools overlap with Google's top-10 organic results, according to secondary sources. Different assistants also cite different sources.
Can I measure AI Overview traffic in Search Console?
AI Overviews and AI Mode traffic is counted under the Web search type in the Performance report and is not reported separately, per Google's documentation.
What you get:
- What Google documents about AI Overviews and AI Mode
- The largest published AI-citation sample, with a per-model breakdown
- Why assistants cite different sources
- How far to trust vendor-produced numbers
- A practical, low-cost action list