82% of Cambridgeshire firms are invisible to AI search.

On 27 September 2026 I asked ChatGPT, Gemini and Google AI Mode for the best accountants, solicitors, recruitment agencies, IT support companies, architects and estate agents in Cambridgeshire: 4 ways a buyer phrases it, 10 times each, per engine. Then I matched every firm named against every firm Google Maps lists in the county. Of 1,826 firms, 328 were recommended at least once. 1498 — 82% — were never named, by any engine, in any of the 720 answers.

Sector by sector

What the named firms have in common

I went through the top and bottom of every table by hand. Three things separate the firms that come up from the ones that never do, and none of them is the size of the firm.

• Review volume, not rating. Named firms average 17, 65, 87, 13, 9, 209 Google reviews across the six sectors; the unnamed ones average 3, 17, 17, 5, 4, 45. 49% of named listings have twenty reviews or more. Of the rest, 15% do.
• A listing in the directory the engine trusts. ChatGPT reads GetAgent for estate agents, the Legal 500 for solicitors, Agency Central for recruiters and the ICAEW register for accountants. Gemini reads Clutch for IT firms. A firm missing from the one directory for its sector is missing from the answer.
• A website and a claimed profile. 94% of named Cambridgeshire listings have a website and 85% have claimed their Google Business Profile. Among the listings never named, it is 68% and 58%. An engine cannot recommend a firm it cannot check.

Those three things are the job. Reviews, the right directory, and a first sentence with a place in it are what a month of local SEO goes on, and none of them needs a bigger firm.

The directories that decide

Every engine cites what it read. Summed across all 6 sectors, these are the sources behind the answers. Google’s own Maps links are left out of the table — AI Mode and Gemini lean on them constantly, which is the point.

The six sectors

Each block is 120 answers: 4 phrasings, 10 repeats, three engines. “Firms named” counts a firm once however many times it came up. The bars are the share of answers in which the firm was named.

Method

• The denominator. Every Google Maps listing in each sector’s categories within the county, pulled through DataForSEO’s Business Listings data, filtered to Cambridgeshire postcodes, wrong categories and permanently closed listings removed. Firms with more than one office are grouped by website, so Sharman Quinney is one firm, not several.
• The prompts. “Best X in Cambridgeshire”, “recommend an X inCambridgeshire”, “X near Cambridge”, “who are the top X in Cambridgeshire”. Each run 10 times per engine, logged out, UK location, because the answer changes between runs and one run proves nothing.
• The engines. ChatGPT, Gemini and Google AI Mode, collected through DataForSEO’s scrapers on the date above. ChatGPT with web search on.
• A mention is the firm’s name, or its website, in the answer. Being cited as a source is counted separately. Matching was automatic, then checked by hand at the top and bottom of every table.
• What it cannot tell you. It is one day’s snapshot and the engines move. Firms without a Google listing are not in the count. The list of firms that werenot named is deliberately not published.

The per-firm data is available to journalists and trade bodies on request:hello@1weeksites.com. If you cite the study, link to this page. I will re-run it quarterly and keep the URL.

The same study, same sectors and same questions, for Suffolk, Norfolk and Essex, and East Anglia as a whole.