Zoe review: £300 of science, £300 of vibes
The test kit behind Zoe is real research with real publications. The app that arrives afterwards turns all of it into a single food score, and can rarely tell you why that score changed. Ten weeks in, we wanted the study, not the game.
Zoe's onboarding tests are genuine research; the app that follows can't explain itself. Consider it if you want a structured, evidence-led starting point and will cancel after the test phase. Skip it if you already log food, or if you want your own glucose curves — there's no raw CGM export.
Wait for v4 — or buy the science without the subscription.
Zoe's onboarding science is the most rigorous in the category. What follows is a gamified food log whose central number is derived, not measured, and which quietly stops being personalised once your sensor comes off. Good intentions, thin second act.
The score, axis by axis
What we liked
- Onboarding tests are genuine, peer-reviewed methodology
- Food database is unusually good on UK and EU products
- The habit nudges are gentle and skippable
What we didn't
- The headline score is modelled, not measured, after week two
- No raw CGM export; you can't audit your own curves
- Two testers, same meal, wildly different scores, no explanation
- £600+ in year one for a food log at the end of it
The two products inside Zoe
Product one is a two-week measurement programme: a CGM, a blood draw, a gut microbiome sample and a standardised meal challenge. It's careful work, and the published methodology holds up better than anything else sold direct to consumers this year.
Product two is the app you live in for the following eleven months. It scores foods, celebrates streaks, and keeps a leaderboard of your week. It is competent and pleasant and, crucially, it stops measuring anything the day the sensor comes off.
You pay for a study and graduate into a game. The disappointment isn't that the game is bad — it's that the study was so good.
The transparency problem
We fed both testers the same six meals in the same order. Scores diverged by up to 34 points, which is exactly what a personalised system should do — but neither app could explain the divergence beyond "your body responds differently." Personalisation without explanation is indistinguishable from noise, and it's why Data rights and Daily UX both land in the fives and sixes.
Who should buy it
- Consider if you want a structured, evidence-led starting point and you'll cancel after the test phase.
- Skip if you already log food, or if you want your own glucose curves out of the system.
Sources
- Zoe — programme structure and pricing. Accessed 1 Sep 2026
- Use of continuous glucose monitoring in non-diabetic individuals for cardiovascular prevention: a systematic review. Cureus, 2025 — 7 studies, ~1,127 participants. Accessed 1 Sep 2026