How to Actually Read a Review
(7 minute read)
The reviews said the hike was easy.
That's what we knew about Capu Rossu in Corsica in August 2022: a headland on the west coast, an old watchtower, and a comfortable consensus online that the walk was no trouble at all. So we parked at the trailhead at eleven thirty in the morning, in August, with one bottle of water each. Four hours later we were out of water and unsure about our futures.
The reviews weren't fake. That's the part worth sitting with. Nobody was paid, nobody was astroturfing, and the trail is, in fairness, easy. It's easy in May. It's easy at seven in the morning. It's easy if you're twenty-four and carrying three liters of water. Every one of those people told the truth, and the truth nearly cooked us.
So here's where we've landed, and it isn't where you'd expect a travel agency to land: we still use reviews. All of them. Every day. The interesting question was never whether to trust them. It's how to read them.
The Number Is Not the Point. The Number of Numbers Is.
Start with the arithmetic, which almost nobody does. A 4.9 with 60 reviews tells you very little, because who's to say the owner doesn't have forty friends and a large family? That isn't fraud, exactly. It just isn't evidence. It's a rumor with a decimal point.
A 4.6 with two thousand reviews is a different object entirely. It has survived contact with two thousand strangers, including the difficult ones, and it's still standing at 4.6. That number has been stress-tested, and the 4.9 has not. There's no clean threshold here and we're not going to invent one for you, because it's a gradient rather than a line. But the instinct to sort by rating, which is exactly what the interface invites you to do, is the single most common mistake people make. It puts the untested 4.9 on top of the battle-hardened 4.6 and tells you it's better.
Read the Bad Reviews. Look for a Throughline.
When a place has two thousand reviews and a strong average, we go straight to the one-star reviews. Not to be scared off, but to look for a throughline. What you'll usually find is a scattering of individual grievances that have nothing to do with each other. Somebody thought the server was rude, which in France frequently means the server was French. Somebody's order came out wrong, and then it was fixed, and they wrote about it anyway. These are not data. These are people having a bad Tuesday, and they tell you nothing about whether you'll have a good one. But if three separate people, on three separate visits, describe the same specific thing, that isn't a bad Tuesday. That's a pattern, and patterns are the only thing in the bad reviews worth your attention.
The corollary is more useful still. A two-star review is often a compliment in disguise, and you can see why if you think about what it takes to leave one. Somebody was angry enough to come home and post about it, but something stopped them short of the full one star, and that something is usually the food. They were furious about the service and the product was too good to condemn. In a coffee shop especially, where service barely matters and the coffee entirely does, a wall of two-star reviews complaining about the staff is close to a green flag.
The anger tells you what the reviewer cares about. It doesn't tell you what you should.
Hotels Are Easier Than You Think
Everybody assumes hotels are the hard case, because a hotel is service all the way down and there's no dish to hide behind. We've come around to the opposite view.
A hotel has less to get wrong. It must be clean and it must be comfortable, and beyond that you're into luxury, which is a preference rather than a standard. A restaurant has to be right about forty things on a Tuesday night in October. A hotel has to be right about two.
And the nicer a hotel is, the less you have to rely on the star average at all, because it will have accumulated other evidence: articles written about it, photographs that aren't its own, a reputation that exists off-platform. That's the cross-reference you want, and it's exactly what a budget hotel lacks. Nobody writes a glowing review of a place that was merely sufficient. They just check out. So the cheaper the property, the more negative-skewed the reviews, and the less the average means.
Beyond that, we do the unglamorous things. We look hard at the photographs, especially the guest photographs, which are the only ones nobody art-directed. And we call.
Where the Whole Method Falls Apart
Now the honest part. Take a village in southern Burgundy, where there are three restaurants, and the best one has eleven reviews, four of which are in French and one of which is about parking. There is no signal to extract and no arithmetic to do. The sample is too small, the reviewers are too few, and the algorithm has nothing to work with and will happily give you an answer anyway.
This is the exact place where the internet stops being useful, and it happens to be the exact place worth going. The star rating is a tool for crowded places, and the more crowded the place, the better it works. Which means it is at its most reliable precisely where you least need it, and useless where you need it most.
What Strangers Are For
Which brings us to the other source, and to the thing reviews structurally cannot do.
Some of the best things that have ever happened to us came from a stranger. A shopkeeper in Paris locked up his store near midnight and walked us to a club. In Marseille, the couple next to us at a bar sent us to another one, and I've been back several times since. Aurélien, who ran the only bar in a Corsican village of a dozen houses, made suggestions and we took them. And we once found ourselves at a pelote basque match in a village we had been about to leave, entirely because somebody mentioned it was happening.
Look at that last one, because it's the important one. There is no review of a thing that hasn't happened yet. The match tonight, the fête on Saturday, the fact that the good baker is closed this week. None of it is searchable. Some of it isn't even a business. It exists only in the heads of the people standing around you, and it evaporates by Tuesday.
That's the gap, and it's permanent. Reviews are a record of the past, and they're a decent one. But the thing you'll tell people about when you get home is very often happening this week, and the only way to find out is to be somewhere with nothing scheduled, talking to somebody who lives there.
So What Did We Do Wrong on the Hike?
We had the reviews. We read them. They were accurate. We also had something better, and we ignored it: that morning, in Aurélien's bar, a local heard our plans and told us in French that it was too hot, and everyone else in the bar agreed with him. We thanked him, and then we went anyway. He had the one piece of information no review could ever contain, because it wasn't about the trail at all. It was about that day, and about us.
That's the part that keeps us humble about all of this. Read the reviews, do the arithmetic, look for the throughline. And then find somebody who lives there, and believe them the first time.
We want the counterexample. What's the best thing a total stranger ever sent you to, and would you have found it any other way? Or tell us we've got this backward, and that you've never once regretted trusting the stars.