Sign inJoin the waitlist

What ChatGPT and Gemini cite for United States buyers: 204 readings, and why the American set belongs to documentation

The US cut of our 15 October 2026 benchmark: 204 readings, zero citations of our own domain, platform documentation at the top of the list, and review directories at 0.8% of all citations.

Published by AI Knows Us · Updated 16 October 2026

In the United States cut of our 15 October 2026 benchmark, ChatGPT and Gemini answered 102 buying questions each, which is 204 US readings, and cited our own domain zero times on either engine. What they cited instead was overwhelmingly somebody's own website: across the full run of 597 readings and 10,652 citations, 85.6% went to a page on the vendor's or the agency's own site, official platform documentation took 9.7%, and review directories such as G2 and Capterra took 0.8%. For a US buyer or publisher, that source mix is the finding, not the rankings inside it.

Those 204 readings are one market out of three. The rest of the run, both engines and all 597 readings, is published as what AI assistants actually cite across 597 readings of ChatGPT and Gemini in three markets. This page is the American slice and what it means if you sell in the United States.

What the US cut covered

Date run15 October 2026
EnginesChatGPT and Gemini
Questions102, of which 72 are worded identically in all three markets and 30 are US specific
US readings204
Readings citing aiknowsus.com0 on ChatGPT, 0 on Gemini
Market controlfixed in the prompt text and verified in the answer, not by an API parameter

Two engines, named. Perplexity, Claude and Google AI Overviews were not in this run, so no sentence here covers them. Our product measures five surfaces and this dataset covers two, and that distinction is worth holding onto when you compare any two visibility numbers.

The domains that keep showing up

These are counts of readings citing each domain across the whole run rather than US only totals, and the top of the list is where the American questions concentrate hardest, because so many of them are about how the engines themselves work.

DomainReadings citing it
developers.google.com178
support.google.com138
blog.google125
help.openai.com112
semrush.com95
tryprofound.com61
otterly.ai58
depra.ai53
peec.ai49
g2.com49

Four of the top five are platform documentation. Ask either engine how Gemini chooses sources, or what a crawler directive does, and Google's and OpenAI's own pages answer it. That part of the US question set is not a competition, and a page written to beat developers.google.com on Google's own mechanics is budget spent for sport.

The specialist vendors, by measured citation count

Below the documentation sits a short list of specialist platforms, counted neutrally. Profound at 61, Otterly at 58, Depra at 53, Peec at 49, Scrunch at 11 and Rankscale at 3. These are the companies whose pages the engines reached for when asked what to use and who to hire. None of them is enormous. They are cited because they published on the subject early and in detail, and that is the only barrier between their counts and anyone else's.

The source mix a US page is competing inside

Source typeShare of 10,652 citations
Vendor's or agency's own website85.6%
News or large publisher2.5%
Official platform documentation9.7%
Review directory such as G2 or Capterra0.8%
Social or community0.5%

Read the bottom two rows before the top one. The standard American advice for a software or services company is to build a review profile and wait for the recommendations to follow. In this run, review directories carried 0.8% of citations. One directory domain, g2.com, does appear in the top ten at 49 readings, which tells you the whole category is close to being one site, and that site is not what the engines reach for when they answer a buying question.

The top row is the better news. 85.6% of citations went to self published pages. A US company competing for those citations is competing against other people's marketing pages, which is a fight that rewards evidence, dates and specifics rather than domain size.

What is missing from the American list, and why it matters

The locally specific winners in this run are not American. The domains cited dozens of times on the strength of being local to one market are British and Indian: whito.co.uk, rank4ai.co.uk and aeoscore.co.uk in the United Kingdom, and themediaverse.in, upgrowth.in, nicodigital.com and hawkorbit.com in India. No US only domain appears among them.

So the American set is a harder market to win on geography and an easier one to win on evidence. Being based in Denver earns a US company nothing here. Publishing a dated measurement with a sample size, on a question American buyers actually ask, puts it in the 85.6% column where the citations already are. The same discipline that makes a search vendor's claims checkable is worth applying when you buy: the questions in running an SEO procurement you can defend work as well on an AI visibility pitch as on a search one.

Questions American buyers ask about these 204 readings

Why are the domain counts run wide rather than US only?

Because the counts we publish are readings citing a domain across all 597 readings. The US specific figures on this page are the ones labeled as such: 204 readings, 102 questions, and zero citations of our own domain on either engine.

Should a US software company still build a G2 profile?

For selling to people, yes, because American buyers read those listings. As a route to being cited by ChatGPT or Gemini, the whole directory category came to 0.8% of citations in this run, so it belongs in the sales budget rather than the citation budget.

Which US questions did the engines answer with documentation?

The technical ones. How an engine selects sources, what a crawler directive does, how a feed or a markup choice is read. Google and OpenAI pages took those, which is why the top four domains in the table are theirs.

Can I reproduce the US cut myself?

Yes, and the method is published for it. Write buying questions that never name your company, lock the market in the prompt text rather than in a location parameter, run them on ChatGPT and Gemini, and record every cited URL. The market lock and the rest of the protocol are in our published measurement method.

When does the US cut get run again?

The same 102 questions go back to both engines every 60 days, and each round is published with the date it was taken.

What a US publisher should take from 204 readings

Concede the documentation questions. Compete where the 85.6% lives, which is your own site, with pages that carry a date, a sample size and the engines named. Treat a directory profile as a sales asset rather than a citation strategy, on the strength of 0.8%. And keep your own count: 204 readings produced zero citations of our domain, which is the number we have to beat and is the shape of the number you should be holding too.

Disclosure, because this page is measurement published by an interested party: AI Knows Us sells AI visibility measurement and the publishing work that follows it, across five surfaces including the two counted here. The US figures above are from our own run, and the method is published so the result can be checked without taking our word for it.

Read next

AI Visibility Audit Deliverables for a UK Agency: The Five Artefacts, and Why the Zero Column Is One of ThemWhat a British agency should hand a client at the end of an audit, including the findings that did not survive a second reading.A Workflow Standard for AI Search Optimisation: Six Stations, and Who Owns Each OneThe execution sequence an Indian agency can hold a team to, measured against what our own site serves eleven named crawlers.The Multi Engine AI Citation Audit Framework: One Column Per Engine, Never a Blended ScoreA reproducible audit design for an Indian agency, built from 1,260 readings on 16 October 2026 and the 597 that came eleven days before it.A Small UK Agency Against a Large Consultancy on AI Search Optimisation: What the Citation Data Says Each One Can OwnThree British agency domains took dozens of citations each in our readings, and none of them is a large firm. Here is what that result does and does not tell a ten person shop.Build or Buy an AI Visibility Stack: The Five Silent Wrong Answers an In House Build ProducesWhat it actually takes to run a multi engine tracker yourself, measured from the defects in our own build, with rupee prices for the alternative.AI Search Optimisation and SEO in India: What Each One Costs, and What Each Return Is Counted InMost of the production work is already inside your search invoice, two budget lines are genuinely new, and the two results are counted in units that cannot be divided by each other.

Or browse the whole archive.

See what AI says about you.

The first scan is free, it needs no account, and it asks the questions your buyers ask.

Free. No card. We ask 5 real buyer questions on 2 AI apps.

Already sure you want this? Join the waitlist. We open on 31 October 2026. It costs an email address. No card, and nothing is charged. Or talk to a person if you have a question first. Or read how we fix what it finds.