Why we don't call it "AI-powered pricing"

Hos91 suggests room rates based on your own booking history. We could call that "AI-powered revenue management" - plenty of hotel software does, and it would fit right in on this page. We're not going to, because it isn't, and we'd rather you understood exactly what it actually is.

What it actually does

For a given future date, Hos91 looks at how many rooms are already booked for that date and compares it against how many rooms were typically booked at the same lead time - the same number of days out - for comparable past dates on the same weekday. That comparison produces a "pace" number: are bookings for this date running ahead of, in line with, or behind where they usually sit at this point. That pace signal is blended with the room type's own occupancy history to produce an occupancy forecast, and the forecast is mapped onto a fixed set of pricing tiers - book faster than usual, the suggestion moves up; book slower, it moves down.

That's the whole mechanism. It's real arithmetic over real data your own property generated, and every number it produces can be explained and traced back to the actual bookings that drove it. There's no model being trained, no black box, and nothing that behaves differently from one day to the next for reasons nobody can point to.

Why that's the more useful thing, not a lesser one

A rate suggestion you can't explain is a rate suggestion you can't fully trust, especially when it's telling you to raise or drop a price on real revenue. If a suggestion looks wrong, you can actually ask why - and get a real answer, not "the model decided." That matters more at a 20-to-150-room independent hotel, where the person reading the suggestion is often the owner themselves, not a dedicated revenue manager who's used to treating a pricing tool as a trusted black box.

It also means the suggestion is never applied automatically. Every rate change is still a manual action - you see the forecast, the pace signal, and the suggested tier, and you decide whether to apply it. The tool's job is to surface a real, well-reasoned signal faster than you could compute it by hand; the pricing decision itself stays yours.

The same honesty applies to anomaly detection

The same principle runs through Hos91's night-audit anomaly checks - the system flags things like an unusually large discount or a rate that doesn't match what was quoted, using the same kind of explicit, checkable rules rather than a model trained to spot "unusual" patterns it can't fully account for. If it flags something, it can tell you exactly why.

None of this is a claim that deterministic rules will always beat a more sophisticated model - it's a claim about what we're actually shipping today, described accurately instead of dressed up. If you want to see the real pricing calendar and pace signal on your own property's data, book a walkthrough or start a free trial.