How Deal Signals work
Methodology v1.0.0 · last updated 2026-07-10
A Deal Signal is our provisional assessment of one specific fare against comparable recent observations on the same route. It is deliberately not a grade or a certification: our observation history is young, and we would rather show you a small honest sample than a confident-sounding number we cannot support.
The rules we hold ourselves to
- Like-for-like only. A fare is compared exclusively against observations with the same passenger basis, trip type, cabin, stops class, market, currency and similar trip duration. A nonstop fare is never judged against connecting fares.
- Minimum sample. A signal is computed only with at least 3comparable observations. Below that, we say “not enough data yet” — we never extrapolate.
- Notable currently means at least 15% below the median of comparables. This threshold is a calibration parameter and will be tuned openly as our data grows; the version number on each signal tracks this.
- Everything is visible. Each signal shows its sample size, coverage window and data-source categories. Prices themselves always carry their observation time.
- Commission never touches assessments. Signals are computed purely from price evidence. Which partner pays what plays no role — structurally, the signal engine has no access to commercial data.
Why “provisional”?
A trustworthy price-history assessment needs long observation coverage. Ours is accumulating — signals will graduate to fuller assessments only when the underlying history genuinely supports it, and this page will document that change when it happens.