Booking.com review score: how it is calculated and what moves it

The Booking.com review score is a 1 to 10 number given by guests, not by Booking.com. It is built from guests' overall scores over the last three years, and since January 2025 more recent reviews carry more weight. Optional subscores such as cleanliness and staff do not count toward it.

The score sits open in a browser tab from breakfast to night audit. It moves by a tenth and nobody in the building can say which review did it, or why last month's good ones did not cancel this week's bad one. After reading this you will be able to say what the number is built from, which of your departments owns each part of it, and what to do in the next week rather than the next quarter.

What is the Booking.com review score, and what are the stars and yellow squares?#

The review score is a 1 to 10 number that guests assign, not a judgement Booking.com makes about your hotel. Two other symbols sit next to it, and they mean different things.

The score appears as a white 1 to 10 number in a blue square. It is the aggregate of what guests scored, over the last three years. The yellow stars, 1 to 5, are the star rating: Booking.com assigns it from its own assessment of the property's facilities and services, so it describes the class of the property, not how guests felt about the stay. The yellow squares, 1 to 5, are the quality rating, which Booking.com also assigns, as its own description of star ratings, review scores and quality ratings (opens in a new tab) explains (updated 31 May 2025).

The score is not one review and not a lifetime average. It is a moving window, and the window is weighted, which the next sections unpack. Subscores are a separate layer: guests may rate location, cleanliness, staff, comfort, facilities, value for money and free Wi-Fi, each submitted independently of the overall score.

The practical consequence: a traveler comparing two hotels is reading three different signals at once. When you audit your own page, read them separately instead of collapsing them into one number.

Who can leave a review, and when?#

Only a guest who booked the stay through Booking.com and either stayed or arrived without staying can review it, and the window closes.

Eligibility covers two cases: the guest stayed, or the guest arrived but did not actually stay. A guest who never arrived is not in this group. After check-out, Booking.com emails the guest, who then has 90 days to leave a review, according to Booking.com's partner help article on Guest Review Scores (opens in a new tab) (read 7 November 2026). The public "How we work" page says Booking.com may accept only reviews submitted within 3 months of check-out; treat 90 days as the working figure.

Reviews stay online for 36 months. After that they are archived and the overall score is recalculated from the reviews still showing. Booking.com may also stop showing reviews when the accommodation changes ownership.

Fake reviews are detected by people and automated systems and deleted when found. Anyone can report a suspicious review to Customer Service, and the fraud team investigates. Your hotel may reply to a review. It cannot delete, edit or select which reviews appear.

The operational reading: the guest who had the problem is the one who receives the email, so recovery has to happen before check-out, not after. If the air conditioning failed on Tuesday night, the conversation that prevents a 4 has to happen on Tuesday night.

How does the weighted score work, and why can one bad month pull it down?#

Since January 2025 the overall score is a weighted calculation in which recent reviews have a bigger impact and a review counts for less as it gets older. Before that it was a flat average of the last three years.

Booking.com publishes the direction, not the formula. There are no published weights and no published decay curve, so do not rebuild it in a spreadsheet and present the output as fact. One published detail matters for smaller properties: where a property has few reviews, the most recent reviews are given the same weight, to limit large swings. A small review base therefore behaves differently from a large one.

The mechanism is simple to state. Under a flat average, a bad month is diluted by years of good ones. Under recency weighting, the same bad month sits closer to the top of the stack and pulls harder. The effect works in both directions: a run of good recent reviews lifts the score faster than it would have before 2025, which is the part hoteliers usually miss.

What this means in practice is that recent reviews describe your score better than older ones do. The number is a report on the recent operation, not a verdict on the hotel. The 36-month archive matters here too: when old reviews drop out, the base the score is calculated from changes even if no new review arrives.

What do the subscores point to in your hotel?#

Each subscore names a part of the operation, which is why they are more useful to you internally than the overall number is. Remember that they are optional for the guest and do not count toward the overall score, but guests still read them on your page.

Cleanliness points at housekeeping: room checklists, bathroom and corridor standards, linen timing. Staff points at the front office and everyone else the guest meets: check-in wait, problem handling, whether a request was followed up. Comfort points at rooms and maintenance: the bed, air conditioning, noise, blackout, water pressure. Facilities points at what exists and whether it works on the day: lift, pool, gym, parking.

Value for money is the revenue manager's lever. It compares the price the guest paid with what was delivered, so a rate rise with no change in delivery moves this one and leaves the others alone. Location is the one subscore the hotel cannot change; spend the effort on the ones you can move. Wi-Fi and breakfast are optional ratings and do not count toward the overall score, but guests read them, so they still shape the decision.

One caveat: because subscores are optional, a single subscore can rest on fewer responses than the overall score. Read the count, not just the average.

Why does the score fall? Three patterns in your own reviews#

A falling score is almost always a repeated cause, not one angry guest. Read the last 20 reviews by date and count what recurs. (If you want the wider picture of how review scores fit with other guest satisfaction measures, see guest satisfaction metrics explained: review scores, NPS and complaint rates.)

Pattern one, a physical defect: air conditioning, hot water, the lift, street noise. It appears in different words but the same noun. The fix is a maintenance ticket with a date, not a reply.

Pattern two, a service break at one touchpoint: the 15:00 check-in queue, breakfast not replenished at 09:30, a transfer that arrives late. Reviews usually carry the time, which is what makes this pattern findable.

Pattern three, an expectation gap: the photos, the room description, the sea view that is a side view. This is a distribution problem, and the operational team cannot fix it by working harder.

Pattern four, a changed comparison set: a new property opens nearby at the same rate. Nothing in your building changed; the guest's reference point did. There is also a structural reason: recency weighting means a strong period from two years ago stops protecting the current score as new reviews arrive.

Read by date, not by score. Reading only the 1s tells you what went wrong; reading all of them tells you how often, which is the number that decides where to spend.

What can the hotel do about the score?#

Fix the repeated cause, ask at the right moment, and reply in public. In that order, because only the first one changes what future guests write.

Fix first. A reply that says the compressor was replaced is worth more than an apology, and it is only worth anything if it is true. Ask every guest, not only the happy ones: selecting who gets asked, or offering something in exchange for a score, misleads you as much as the guest, and it is the kind of practice platforms act on; confirm the current rules in the partner help pages. The mechanics of asking without breaking them are worth studying separately; see how to ask hotel guests for reviews without breaking the rules.

Reply to reviews: short, factual, about the mechanism. The reply is read by the next guest, not by the last one, so write it for the person still deciding. Do not run the same template under every review; guests recognise it and discount it.

Put the subscores on a monthly agenda by department. The overall number is the output; the subscores are the inputs you can assign to a person. Do not try to buy the number back with a discount after the stay: by then the review is already written. Keep the review work separate from the sales work. A score is an output of operations, and treating it as a campaign produces short-term movement and long-term noise.

Does a higher score bring more bookings, and how does it connect to ranking?#

No published rule converts a score into bookings or into a position on the results page. The score is one signal among several, and the ranking formula is not public.

Booking.com does not publish how the results list is ordered. Treat any factor list you read online, including this one, as unverified, and check the partner help pages for the current description. The ordering is not published, so any factor list is unverified; check the partner help pages.

Read any partner program's current conditions on the partner side rather than modelling eligibility from your score. The practical read: the score is a slow signal you influence through operations, not a switch you flip before high season. If you want to understand placement, study it as a separate problem from review quality, because the levers are different; see ranking on booking sites: what moves a hotel up the list.

What to do next#

Take the last 20 reviews tonight, tag each one with a pattern, and open a single ticket with a name and a date on it. The score is weighted toward the recent past, so it reports on the operation you are running now. The subscores tell you which department owns the cause, the recency weighting tells you how quickly new reviews will show whether the fix worked, and the reply tells the next guest that it was fixed.