Without web search, Bupa comes first in half of all answers, and one assistant prefers Aviva on four of eight questions.
Measured without web search. The search run comes in the next edition.
United Kingdom
Bupa is AI's answer to private healthcare in Britain. One assistant disagrees.
Measured without web search. The search run comes in the next edition.
The brands AI names first most often, plus any this finding is about.
| Brand | Named first | Named at all |
|---|---|---|
| Bupa | 86% | |
| Aviva | 72% | |
| Vitality | 68% | |
| AXA Health | 84% | |
| WPA | 58% | |
| Saga | 5% | |
| Freedom Health Insurance | 10% | |
| Benenden Health | 4% | |
| Simplyhealth | 3% | |
| Medicash | 2% | |
| Westfield Health | 3% | |
| The Exeter | 22% | |
| HSF Health Plan | <1% | |
| BHSF | <1% | |
| Cigna | <1% | |
| Health Shield | <1% | |
| National Friendly | <1% | |
| None of the tracked brands |
720 answers across ChatGPT, Gemini and Claude together, October 2026. These are AI answers, not market shares. The three assistants share a favorite on 3 of 8 questions.
By assistant, and on your own buyers' questions: that is a Position Map. See a Position Map
Places count every brand AI names, so the numbers can jump. "=" is a tie. "–" means AI never names it first.
The leader is clear, but not secure. Ask which insurer pays its claims, and one assistant is split between WPA and Bupa while another sends people to Aviva. In a purchase built on trust, the insurer a family hears first now depends on the assistant they asked.
BupaThe assistants differ.
Answers vary from one ask to the next, and the apps can search the web and remember the person asking. That is why we ask each question 90 times without web search and 36 times with it.
BupaThe assistants differ.
BupaThe assistants differ.
BupaAll three agree.
BupaThe assistants differ.
BupaAll three agree.
BupaAll three agree.
BupaThe assistants differ.
Is your company missing or measured under the wrong name? Tell us. We correct facts at any time. Results change only in the next edition, for everyone.
Edition 1, Q4 2026. Next edition: January 2027.
Eight buying questions per market, each asked in six wordings of ChatGPT, Gemini and Claude: five times without web search (720 answers per market, 14,400 in all) and twice with it (288 per market, 4,600 in all, in sixteen of the twenty markets so far). Measured 25 to 30 October 2026. Kesterley.
Your own questions, not ours. Position Map: where you stand on the questions your buyers ask, why AI puts a rival first, and what to do first. From $2,500. Two weeks.
Talk to us