Elementera
Products
OriginumLenzThe Supervised plan, our managed AEO service
Case StudiesBlogFree ToolsResearchAbout UsBook a call
Book a call
  1. Home
  2. Claims we tested
  3. Claim

Claim: check your AI visibility once and you know where you stand

By Mojtaba Ebadi, Co-Founder and Chief Operating Officer. Reviewed by Dr. Maziar Gomrokchi, Co-Founder and CEO. Last updated October 22, 2026.

Key point. Not supported. When we asked the same questions again, the cited websites changed on most of them, on Google AI Overviews and on ChatGPT. One check is a snapshot of one moment, not a measurement of where you stand.

Verdict: not supported. A single screenshot of an AI answer tells you what that answer said once. Ask again and the sources, and sometimes the answer itself, change.

What did we measure?

We asked the same questions more than once and compared the websites each answer cited. We did this on Google AI Overviews and, separately, on ChatGPT.

14 of 118
Google searches that returned the exact same set of cited websites on all three checks over two days (no interval published for this count)
Elementera, Google AI Overviews, US non-branded business searches, July 2026

On Google AI Overviews: "When we checked the same 118 searches three times over two days, only 14 of them returned the exact same set of cited websites each time."

Even whether an AI Overview appears moves: "Whether an AI Overview appears at all also moves. In the same three checks over two days, 22 of 150 searches showed an AI Overview on some checks and not on others (15%; 95% interval 10% to 21%), so a KPI that counts AI Overview appearances can change with nothing changed on your side."

On ChatGPT: "Asking ChatGPT the same question three times in a row, the sets of websites it cited overlapped by only about 30% on average, across 100 non-branded questions." That 30% is an overlap index: the websites two answers share, divided by all the websites either one cited. It is not a share of citations.

Method

The table gives the engine, the samples, the intervals and the dates behind each number above.

ItemDetail
EnginesGoogle AI Overviews, US English desktop; ChatGPT, read from its interface through a third-party service. Reported separately, never pooled
Samples118 searches, 3 checks (same sources); 150 searches, 3 checks (AI Overview shown or not); 100 questions, 3 asks each, 3,600 pairs of answers (ChatGPT overlap)
95% intervals10% to 21% for the 22 of 150; none published for the 14 of 118 or the ChatGPT overlap
DatesJuly 16 to 17, 2026 (Google); ChatGPT finding registered July 16, 2026
DesignRepeat measurement of the same questions with nothing changed on any site's side

Each number has a methodology card. The full discussion is in why most AI visibility numbers mislead.

Where does the claim come from, and what do the docs say now?

The claim treats an AI answer as a fixed result. Google described the first version as a "snapshot", and its current documentation says the answers and links vary.

SourceWhat it saysPublishedRead
Launch announcement: Google, Supercharging Search with generative AI"You'll see an AI-powered snapshot of key information to consider, with links to dig deeper."May 10, 2023October 22, 2026
Current documentation: Google Search Central, AI features and your websiteAI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary.current pageOctober 22, 2026
The common claim"Check your AI visibility once and you know where you stand."repeated widelyOctober 22, 2026

What the common claim gets wrong. A snapshot is one moment. Google itself says the responses and links vary, so a reading needs the same questions asked more than once before it describes where you stand.

How far does the result reach?

The Google results cover US non-branded business searches; the ChatGPT result covers 100 non-branded questions read from the ChatGPT interface. Neither says how much any other engine moves, and the two are not combined into one figure.

What carries across is the lesson: any engine whose answers move needs repeated checks before a change can be read as real.

What should you do instead?

Treat a single check as a starting point, and build a measurement around it.

  1. Fix a list of buyer questions and keep it the same from one check to the next.
  2. Ask each question at least three times, on more than one day.
  3. Record citations and mentions per engine, never pooled.
  4. Before you call something a change, check that it is bigger than the movement you saw with nothing changed.

Work with us

If you want a team to measure and improve how AI search treats your company, our managed AEO services do that work with you: Book a call. If you would rather see where you stand first, Lenz gives you a free analysis of how ChatGPT and Gemini answer your buyers' questions: Get your free AI visibility report.

More claims we tested.

Elementera
Expertise in. Original content out.

Elementera turns what a business knows into original content for AI search, and measures how AI engines describe and recommend it.

Product
OriginumManaged AEO ServicesLenzBook a Call
Learn
BlogCase StudiesFree ToolsResourcesResearchEvents
Guides
What is AEOAEO vs SEOAI visibility funnelllms.txt guide
Company
About UsContactSite map
hello@elementera.comTrustPrivacy PolicyTerms of Service

Supported by

Google
Microsoft
NVIDIA
Cloudflare
Anthropic

Machine-readable summary: llms.txt ยท Company facts