A couple books a 9.2-rated hotel in Lisbon for four nights in May, arrives to find the “king bed” is two twins pushed together and the “quiet courtyard view” faces a service alley where crates get stacked at 5:40am. The score was real. The reviews behind it were not written by anyone in that couple’s situation, and that is the whole problem with the numbers.
“Sort by newest, and the rating means something”
Mostly true, with a caveat that costs money. Scores decay. A hotel that earned 9.1 in 2023 through new mattresses and a functioning lift can hold that average for two more years while the lift breaks in March and nobody fixes it until autumn, because averages are slow and refits are not. Sorting reviews by date, rather than by “most helpful”, surfaces the last ninety days. On a 600-review property, 90 recent reviews is still a decent sample. On a 40-review guesthouse it is four, and four tells you nothing except that the host replies fast.
The mechanism matters more than the score. Booking platforms verify a review by matching it to a completed reservation, which removes invented reviews but not unrepresentative ones. A person who paid €89 in February grades differently from one who paid €210 during the Champions League final. Filtering by traveller type helps; filtering by room type helps more, and few people do it.
“Photos and detailed reviews beat the star rating”
True, and this is where the real work sits. Three things to check before paying:
- Whether the room in the photos matches the room in the rate you selected, not the category above it
- The date on the most recent review mentioning the lift, the air conditioning or the wifi
- Whether complaints repeat across different guests in different months, which indicates a building problem rather than a bad week
- What the hotel’s replies say, since a defensive response to a fair complaint predicts how a problem gets handled on site
Repeated, dated, specific complaints are the signal. One angry paragraph with no date is noise. A guest who writes “the 4th floor corridor smelled of drains on 12 and 13 March” has given more useful information than forty people writing “great location, would stay again”.
Where the booking actually happens changes what you can see. Direct bookings sometimes expose reviews the platforms never received, and consolidators bundle a rate with terms you cannot inspect until you are standing at reception. This is the sort of gap Kupi Travel was built to close, putting the rate and its conditions in one view so the comparison is between the things that differ rather than the things that look alike.
See also: How Minnesota Businesses Improve Google Rankings
“A high score from any guest is a high score for you”
False, and it is the belief that ruins trips. A 9.4 built on 200 solo business travellers describes a hotel that is efficient, well located for offices, and indifferent to whether two people sharing a room have anywhere to put two suitcases. Filter to couples, or to families, or to whatever you actually are, and the same property can drop to 8.3. The score did not lie. It answered a different question.
Price per person sharpens this. A €180 double is €90 each, which is often cheaper than two hostel beds at €48 in the same city, and the reviews for the hostel will be written by people under 25 travelling alone. That mismatch between reviewer and reader is the single largest source of disappointment in accommodation booking, and it is entirely self-inflicted.
The variable that decides it: whether the guests who wrote the reviews were travelling the way you are, on the dates you are, at the rate you are paying. Get those three to line up and the score finally means what you thought it meant.
