The Whisper Index
Every quarter we run thousands of customer-style questions through ChatGPT, Claude, Gemini and Perplexity across UK small-business sectors, and score two things: how often the assistants name a specific local business (visibility), and how often their answers contain at least one factual error (error rate). Figures shown are illustrative.
| # | Sector | Visibility | Error rate | Movement |
|---|---|---|---|---|
| 1 | Cafés & restaurants Assistants love naming cafés — but a third of answers quote stale menus or hours. | 78 | 31% of answers | Steady |
| 2 | Hair & beauty Strong review coverage keeps salons visible; pricing errors are the common slip. | 71 | 24% of answers | Rising |
| 3 | Florists & gifts Seasonal spikes reward whoever the assistants remember first. | 64 | 18% of answers | Rising |
| 4 | Trades (plumbers, electricians) The highest error rate we track — dead phone numbers and wrong call-out areas everywhere. | 52 | 41% of answers | Steady |
| 5 | Health clinics & physio Retired services and old addresses linger for years in directory data. | 47 | 35% of answers | Falling |
| 6 | Accountants & solicitors Assistants hedge towards national firms unless local content is strong. | 39 | 22% of answers | Steady |
| 7 | Independent retail “Permanently closed” errors hit hardest here — often from a single stale map listing. | 33 | 27% of answers | Falling |
| 8 | Garages & MOT Rarely named individually; when they are, opening hours are usually wrong. | 28 | 38% of answers | Steady |