An 8.4 is often a better hotel than a 9.2. Not always. But often enough that treating the headline score as a ranking will steer you wrong on a regular basis.

Hotel review scores look like a simple average of what guests thought. They are not. They are the output of a weighting system that shifted meaningfully in 2026, built on six sub-scores that most travelers never open, drawn from a guest population that self-selects in predictable ways.

Once you know how the number is assembled, you can read it properly in about fifteen seconds. Here is the whole mechanism.

The score is an average of six numbers, not one

On the largest booking platforms, a guest rates six things after checkout. Cleanliness, comfort, location, facilities, staff and value for money. Each gets a 1 to 10. The guest overall score is the average of those six.

The property public score is then an average of all those guest averages. So a headline 8.4 is a mean of means, and it can be assembled in wildly different ways.

One 8.4 might be a property scoring 8.3 or 8.5 on all six categories. Flat, consistent, no surprises. Another 8.4 might be a 9.6 on location and a 9.4 on staff pulling up a 6.8 on facilities and a 7.1 on value.

Those are not the same hotel. The first is a solid, slightly unremarkable stay. The second is a beautifully placed property with warm staff and a tired building that charges too much for it. Which one you want depends entirely on your trip.

The 2026 change nobody announced loudly

Review scores used to be a flat average of everything collected over the previous 36 months. A stay from thirty months ago counted exactly as much as one from last Tuesday.

That changed in 2026. Recent reviews now carry more weight than old ones. The stated reason is that it lets a score recognise recent service improvements faster, which is true. The practical consequence for travelers is more interesting.

Scores now move faster in both directions. A property that changed hands, renovated, or cut its housekeeping schedule will show that in its score within months rather than years. Volatility went up, and volatility is information.

The read is simple. A score that has been stable for two years is a stable hotel. A score that dropped 0.4 in six months is a hotel where something changed, and the reviews from the last quarter will tell you what.

Why hotels care so much, and why that shapes what you see

The stakes for the property are larger than most guests assume. Properties above 9.0 convert bookings at three to four times the rate of those below 7.0, and platform ranking algorithms lean heavily on score. A sustained half-point improvement can support a 5% to 10% lift in achievable rate.

Which means the score is not just a description of quality. It is an input into price. Two identical hotels on the same street with a 0.5 point gap in score will not be priced the same, and the higher-scoring one is not necessarily 5% better to stay in.

That gap is where value lives. The 8.6 next door to the 9.1 is frequently the same experience at a lower rate, because the market prices the number rather than the room. The same distortion shows up in how hotels use minimum stay rules to signal confidence in a date.

The four things to actually check

Review count, before anything else. A 9.4 from 22 reviews carries almost no information. A single ten-person wedding party can produce that. Below about 100 reviews, treat the score as a rough hint. Above 500, it is reasonably solid.

The lowest sub-score, not the headline. Open the six categories and find the weakest one. That is the thing you will be annoyed about. If the weak category is facilities and you only need a bed near the station, ignore it. If the weak category is cleanliness, walk away regardless of the headline.

Value for money as a price signal. This sub-score is the closest thing to a crowd-sourced verdict on whether the rate is fair. A hotel with 9.0 comfort and 7.2 value is telling you plainly that it is a nice room at too high a price. That is a negotiation signal, not a rejection.

The most recent twenty reviews, by date. Sort by newest and read twenty. Under the new weighting these are the reviews doing the most work on the score anyway, and they are the only ones describing the hotel as it exists now.

Where review scores are systematically wrong

Some biases are consistent enough to correct for.

Business hotels score lower than they deserve. Guests rate them against leisure expectations, and a functional airport property near a motorway junction will never win on location or facilities even when it does its job perfectly. A 7.9 business hotel is often an excellent choice for a one-night stopover, which is the same reasoning we used in our piece on whether airport hotels are worth it.

Small properties score higher than they deserve. Fewer rooms means fewer reviews, and guests who chose a twelve-room guesthouse were already predisposed to like it. Selection bias inflates the number.

Location scores measure the neighbourhood, not the hotel. A 9.5 location score in a city you do not know tells you the hotel is central. Central is not always what you want, and it usually costs more.

New hotels open high and drift down. The first six months of reviews come from a property running at low occupancy with fully staffed service. Scores normally settle 0.3 to 0.5 lower once the hotel is actually full.

Star ratings and review scores measure different things entirely, which we covered in our piece on what hotel star ratings actually mean.

A fifteen-second reading method

Check the review count. Under 100, be sceptical. Open the six sub-scores and find the lowest. Ask whether that category matters for this specific trip. Sort reviews by newest and skim twenty. Compare the value-for-money sub-score against the rate you are being quoted.

That is it. Five steps, and it separates a genuinely good hotel from a well-scored one far more reliably than the headline number ever will.

The rate you are comparing against matters just as much. Best shows the lowest available rate rather than the one that pays the platform best, which means the value-for-money question you are asking about an 8.4 gets a different answer depending on where the price came from.

Common questions

Is an 8.4 hotel score good?

Yes. On a 1 to 10 platform scale, anything above 8 is generally considered good and anything above 9 excellent. An 8.4 with 800 reviews and no sub-score below 7.5 is a more reliable bet than a 9.2 with 40 reviews.

How is a hotel review score calculated?

Guests rate six categories from 1 to 10, being cleanliness, comfort, location, facilities, staff and value for money. Those six are averaged into one guest score, and the property score averages all guest scores from roughly the last 36 months, with recent reviews weighted more heavily since the 2026 change.

Why did a hotel review score suddenly drop?

Under the newer weighting, recent reviews move the score faster than they used to. A drop of 0.3 or more in a few months usually means an operational change such as new ownership, a renovation in progress, or reduced service. The last quarter of reviews will normally say which.

How many reviews does a hotel need for the score to be trustworthy?

Roughly 100 is the floor for a meaningful signal, and 500 or more makes the score reasonably stable. Below 50, a handful of unusual stays can move the number by half a point.

Do higher review scores mean higher prices?

Generally yes. A sustained half-point gain in score supports a 5% to 10% increase in achievable rate, so the score feeds directly into pricing. The practical result is that a well-run 8.6 next door to a 9.1 is often the better value.

Images: Hero by engin akyurt. Laptop comparison by cottonbro studio. Both via Pexels. Hotel corridor by JIP via Wikimedia Commons, CC BY-SA 4.0.