[ DISCOVERY · THE DISPATCH ]

Travel Reviews Are Mostly Fake Now and I Found the One Signal That Still Works

The five-star reviews are written by bots. The one-star reviews are written by competitors. Somewhere in the noise there's still one signal I trust completely — and it's the one nobody can fake.

How to spot fake travel reviews in 2026 has become a survival skill, because the review section — the thing we all relied on for twenty years — is mostly garbage now.

I figured this out properly in a hotel restaurant in, of all places, a perfectly nice town in Italy. The place had a 4.8. Hundreds of reviews. Glowing. “Amazing experience!” “The staff went above and beyond!” “A magical evening!” I sat there eating a frankly mediocre, microwaved-tasting risotto, looking at the empty room, and thought: who wrote all those reviews. Because it wasn’t anyone who’d eaten this risotto.

That was the moment the trust broke for me. And once it breaks you can’t un-see it.

Two things happened at once and they poisoned the well together.

First, AI got good and cheap. Anyone can now generate two hundred fluent, plausible, five-star reviews for their business in an afternoon, each one slightly different, each one warm and generic and utterly hollow. They read fine. They read better than real reviews, which is exactly the problem — real people write awkwardly, mention weird specifics, complain about the parking. AI writes smooth praise about “exceptional service” and “wonderful atmosphere,” over and over, forever.

Second, the dark version: planted negative reviews. Competitors, disgruntled ex-staff, occasionally a guest running a refund extortion play. The one-star pile is now polluted too. So you can’t even trust the bottom of the distribution to be honest.

The rating is cooked from both ends. A 4.8 means nothing. A handful of one-stars means nothing. The signal-to-noise has collapsed.

Here’s the thing that actually helps, the pattern I trained myself to see: fake reviews are too good at writing. They’re fluent and generic and they praise the whole experience in vague, sweeping terms. “Everything was perfect.” “Could not fault a single thing.” “A truly special place.” Notice there’s never a specific. Never a dish name. Never a room number. Never a tiny complaint. Just a warm bath of adjectives.

Real humans are not like this. A real person writes “the shower took forever to get hot but the breakfast pastries were unreal, get there before 9 or they run out of the almond ones.” A real person is mildly annoyed about one thing and weirdly enthusiastic about another. They contradict themselves. They mention the parking, the noise, the guy at the desk who was either lovely or grumpy. Specificity is the fingerprint of a real visit.

So here it is, the thing I actually trust now: specific, mildly-negative, recent detail.

Not the rating. Not the glowing top reviews. Not even the volume. I look for reviews that are recent (last month or two), that name specific things (a dish, a staff member, a room, a smell, a time of day), and that have at least one small genuine complaint mixed in with the praise. That combination is almost impossible to fake at scale, because it requires actually having been there, recently, paying attention.

A review that says “great spot but the upstairs rooms get street noise until midnight, ask for the back” — I trust that completely. Nobody plants that. No bot generates that. It’s too useful, too specific, too honestly annoyed. It’s the texture of a real human who wants to help the next person.

A few more things in my kit:

• Photos by reviewers, not the business. Real guest photos — slightly bad lighting, a half-eaten plate, a messy bed — beat the glossy listing shots every time. If all the photos are perfect and professional, be suspicious.
• Check the reviewer’s history. A profile that’s only ever left one five-star review, ever, for this one place? Bot or paid. A profile with a long messy history of restaurants and hotels in cities all over? Real person.
• Cross-platform sniff test. If a place is 4.9 on one site and 3.6 on another, the high one’s probably gamed. The lower, harder-to-manipulate platform is usually closer to the truth.
• Recency over volume. Two thousand reviews from three years ago tell you about a place that may have changed owners twice since. Twenty real ones from this season tell you about the place that exists now.

I miss the old days, honestly. There was a stretch where you could just read the reviews and basically trust them. That’s gone. But it’s not hopeless — the real humans are still in there, complaining about the shower temperature and raving about the pastries, leaving their specific, annoyed, generous little fingerprints all over the mid-tier reviews.

You just have to ignore the smooth stuff and go find them.

Skip the five-star gush and the one-star rage. Go straight to the grumpy, specific, recent middle — that’s where the only honest people left are still writing.

This article contains affiliate links marked rel="sponsored". We may earn a commission at no extra cost to you.