07 · AI search

Ask an AI tool about Harbour & Key by name and the answer is good. Ask who to sell with, and it names Harbour & Key in 4 of 32 answers.

example-property.co.uk · 12 questions put to ChatGPT, Gemini, Perplexity and Claude, plus Google AI Overviews, measured on 8 July 2026, prepared by Robin Laires.

The short version
What we found
All four AI tools give accurate, positive answers when asked about Harbour & Key by name. On the commercial seller questions it is named in 4 of 32 answers, and Google's AI Overviews cite it on none of the 8 searches where they appeared.
Why it matters
A seller who asks an AI tool who to sell with may be pointed to ESPC, McEwan Fraser or Neilsons. The £600 fixed fee already shows up in a few answers, which suggests the position could be won.
What could help
Fuller profiles on the aggregator sites the engines read, a direct answer on the fees page and schema on key pages could help.
4 of 32
Commercial seller answers naming Harbour & Key (8 questions × 4 engines)
8 of 8
Branded answers clean and positive (2 questions × 4 engines)
0 of 8
Informational answers naming Harbour & Key (2 questions × 4 engines)
0 of 8
Google AI Overviews citing Harbour & Key (shown on 8 of 12 searches)
17 of 32
Commercial answers naming ESPC, the most named rival
Worth protecting

Where Harbour & Key already shows up

Ask ChatGPT or Perplexity “who should I sell my flat with in Edinburgh” and both name Harbour & Key. Ask Gemini for the “fixed fee estate agent in Scotland”, or the “best estate agent in Edinburgh for selling a flat”, and it names Harbour & Key too. The £600 fixed fee and the flat-selling angle look like the most retrievable things about Harbour & Key, and they surface without any of the fixes below in place.

Branded questions (“Harbour & Key Property reviews” and “is Harbour & Key a good estate agent”) come back clean and positive on all four engines. The answers draw on the site itself, allagents.co.uk and Trustpilot, and ChatGPT, Perplexity and Claude cite example-property.co.uk directly.

Method and limits

Measured on 8 July 2026 with the same 12 questions put to ChatGPT, Gemini, Perplexity and Claude, and 12 searches checked for Google AI Overviews. Each engine was run once, and AI answers change from one run to the next, so these counts are a snapshot.

The commercial count covers 8 questions across 4 engines, which makes 32 answers. The branded and informational counts each cover 2 questions across 4 engines, which makes 8 answers.

Real client work. Names and locations changed.