How we decide something is a deal
We check hotel prices the same way every night and compare them with prices we have recorded. A deal has to clear a measured comparison with similar hotels and, once enough history exists, with that hotel’s recorded rates for the same stay configuration. Each deal shows which comparison its discount uses.
What we check
A fixed set of hotels in 107 cities — 13,010 properties in total. For each one we ask the same standardised question: one night and seven nights, for two adults, refundable and non-refundable, at fixed distances ahead of the stay. A property has no single price — only a standardised question does — so we never pretend it has one; we track each of those questions separately.
How often we check it
Every night. Each night’s answer is stored and never edited — if we get something wrong we add a correction rather than rewriting history, so the chart you see is the record we actually collected.
Once a week we also ask what one refundable night for two adults costs at those hotels on dates spread across the coming year. That is what lets us say when a city is cheap, rather than only what it costs today. Monthly comparisons currently cover 68 of the 107 cities we track.
The year-ahead chart is a dated snapshot of forward quotes. Each stay is quoted at a different distance ahead, so the curve reflects both the travel month and how early it is being booked. It is not a year of historical observations.
What counts as a deal
Two comparisons, and a price has to pass both once we know enough. First, against comparable rooms in the same city on the same night — that catches a hotel that is simply cheap for its class. Second, against that hotel’s own recent rates for the same stay configuration — that catches a hotel that is cheap today relative to itself.
Requiring both is what filters out seasons. A room in a Mediterranean city in October looks like a bargain against its summer prices; comparing it with its neighbours on the same night shows it is simply October.
There is also a floor in absolute terms. A large percentage off a cheap room is not worth anyone’s attention.
Before we have 10 recorded prices for a stay configuration, a deal is compared with similar hotels only. Its page names that comparison and the number of hotels behind it. Once that history exists, both comparisons must pass and the published discount uses that configuration’s recorded median, with its sample size shown.
What happens before publication
Selection and publication are automatic. The system rechecks the stay’s price and availability with our rates supplier before publishing and records when that check happened. Prices can change afterward; checkout checks the chosen room, price and booking terms again before payment.
What we don’t do
We do not predict prices. We can tell you what a room has cost and how much it swings; we cannot tell you what it will do next, and saying otherwise in public would be a guess wearing a number’s clothing.
We keep refundable and non-refundable rates in separate series. We check the supplier’s recorded room name and meal plan against the usual offer for that stay configuration; detected changes block automatic publication. These checks depend on the descriptions the supplier returns. They cannot establish that every recorded price covers an identical room and meal plan.
We track 13,010 properties in 107 cities. If your hotel isn’t one of them, we have nothing useful to tell you about it.
How we make money
We make money on the rooms we sell, not on what we publish. The threshold is a number, not a judgement call.
What the checking produces
One email when a month you pick gets materially cheaper — judged against the same 13,010 properties, checked the same way every night. Nothing in the weeks nothing moves. Email only, no account, no password.
Continue with your account, then review where and when to watch. Weekly deal emails are optional.