Benchmarks & data
These are the underlying tests our rankings are built on — the methods, the meals, and the numbers, written up so you can check our work.
Benchmark protocol
Our fixed-list method for measuring how completely each app's database covers branded, restaurant, fresh and international foods — and how often it falls back to a vague entry.
Benchmark protocol
The stopwatch protocol behind our logging-speed scores — what counts as a completed log, how we average trials, and why speed predicts whether people keep tracking.
Benchmark protocol
The full method, sampling and error math behind our headline AI calorie-accuracy benchmark — how we build the reference meal set from our 12,000-user study and compute MAPE.