What "26% below typical" means, and what it deliberately doesn't
Most price claims compare a fare against an average that includes flights nothing like yours. Ours compares against the same route, the same month and the same booking window — and nothing else.
When one of our alerts says a fare is 26% below typical, that sentence is doing precise work, and it is worth knowing exactly what it does and does not claim.
What it compares against
The comparison is against fares we have recorded for:
- the same route — not the region, not the country pair
- the same departure month — because seasonality is the largest single effect in most fares
- the same booking window — how far ahead of departure the fare was seen
That is the whole comparison set. If we have watched JFK to Lisbon for five weeks, "typical" means the median of what we saw for March departures booked roughly this far out. It does not mean the median of all transatlantic fares, or of that route across all seasons, or of what anyone else has published.
Why the narrower comparison is the honest one
A broader baseline makes almost every fare look like a deal, which is precisely why broader baselines are popular.
Compare a March fare against a full-year average and you will find it "below typical" whenever March happens to be cheap on that route — which tells you nothing about whether this particular March fare is good. Compare a fare booked three months out against fares booked three days out and everything looks like a bargain.
Both of those produce impressive percentages. Neither answers the question you actually asked, which is: should I book this now, or wait?
What the number does not tell you
It is not a prediction. We will say what a route has typically done; we will not tell you what it is going to do next. Nobody can, and a product that pretends otherwise is selling confidence rather than information.
It is also not a comparison against other sellers. We are not claiming this is the cheapest place to buy this fare — only that this fare, on this route, at this moment, sits well below what we have seen before.
The percentage is computed, never written
One detail that matters more than it sounds: every number in the reasoning paragraph of our alerts is calculated by code before the sentence is written. We do use a language model to turn those figures into readable English, but it is handed a fixed set of numbers and every figure it produces is checked back against them before the email is allowed to send.
If a number appears that we did not compute, the sentence is thrown away and rewritten. If the rewrite fails too, you get a plainer sentence assembled from the data directly.
You should never see a number from us that nothing in our database supports.