Media resources
Everything a journalist needs to cover Discoverable or cite our research on how AI assistants describe brands. For interviews, comment or custom data pulls, email hello@bediscoverable.ai. We reply within one working day.
The research: first on Google does not guarantee a mention in ChatGPT
In September 2026 we put the same 1,000 buying questions to Google and to six consumer AI services, three times each, and collected 17,537 completed AI answers alongside 3,000 Google results captures. We then checked whether the business ranking first on Google was named in the AI answer to the same question. ChatGPT did not name it in 72.5% of eligible answers. The full method, results tables, analysis code and worked examples are free to download without registration.
72.5%
of eligible ChatGPT answers did not name the business ranked first on Google for the same question
17,537
completed AI answers collected across six consumer services, in three rounds
1,000
buying questions across 25 commercial categories, fixed before collection
Attribution: “Discoverable, UK AI Shortlist Study, September 2026” with a link to the study page. Data collected 6 to 7 September 2026; analysis dated 14 September 2026.
The research: how often AI gets brands wrong
In July 2026 we asked one AI assistant, Claude, through the Anthropic API rather than the consumer app, to describe 30 real companies, first from its training memory and then with live web search enabled, and scored every claim against each company's own published facts. The full method, results table and limitations are in the study itself, which is free to cite with attribution and a link.
27 of 30
brands were described with at least one materially false claim when one AI assistant (Claude, via the Anthropic API) answered from memory alone (90%). July 2026.
13 of 30
brands were still misdescribed by the same assistant even with live web search switched on (43%). Not a consumer-interface measurement.
900M+
weekly users now ask AI assistants questions before they buy (OpenAI, reported ChatGPT weekly active users, February 2026).
Attribution: “Discoverable, July 2026” with a link to the study. We can re-run the test on a sector or brand list relevant to your story on request.
Boilerplate
Discoverable (also written Discoverable AI, the trading name of Be Discoverable Ltd, London) is a Generative Engine Optimization (GEO) platform that shows brands what AI assistants such as Claude, ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews and Google AI Mode say about them, scores their visibility against named competitors, and gives them a prioritised plan to improve it. Anyone can run a free check at bediscoverable.ai with no login; the free snapshot runs on Claude, and paid runs measure ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews and Google AI Mode too.
Company facts
- Product: Discoverable (bediscoverable.ai)
- Legal entity: Be Discoverable Ltd, Company No. 17383714
- Registered office: 71–75 Shelton Street, Covent Garden, London WC2H 9JQ, United Kingdom
- Founded: 2026, London, United Kingdom
- Media contact: hello@bediscoverable.ai
Logos and screenshots
Our logo mark is below. For high-resolution product screenshots, founder photos or anything else, email hello@bediscoverable.ai and we'll send a full pack the same day.
