You're hungry, you search for restaurants nearby, and you sort by rating. At the top: a place with a perfect 5.0. Sounds great. You go. It's fine, maybe a little sad. Meanwhile the spot down the street with a 4.7 has a line out the door every weekend.
What happened? Look at the review count. That 5.0 probably had four reviews.
Here's the thing about a 5.0: it's not really saying the place is amazing. It's saying not enough people have shown up yet to knock it down.
Think about your own favorite restaurant. The one you'd recommend to anyone. Does it have a perfect rating? Almost certainly not. Somebody got a cold plate once. Somebody waited too long for a table. Somebody was just having a bad day and took it out on the review box. Serve enough people and a few bad reviews are guaranteed.
So a business that's served thousands of customers and still sits at 4.7 has been tested. A business at 5.0 with four reviews hasn't been tested at all. Same little stars, completely different amount of proof behind them.
A high rating with few reviews isn't good news. It's just no news. Check the number in parentheses before you get excited about the stars.
It's the same reason you shouldn't judge a coin by three flips.
Flip a coin three times and getting heads all three times is no big deal — it happens about one time in eight. Flip it three thousand times and getting all heads never happens, ever.
Reviews work the same way. Three glowing reviews in a row? Totally normal, even for a mediocre place. Three thousand glowing reviews in a row? Impossible. Real life gets in the way.
Which means when you sort a list by star rating, the top of that list fills up with the places nobody has been to yet. You're not seeing the best restaurants. You're seeing the least-visited ones.
The trick that IMDb and Amazon both use is simple once you strip the jargon off it. Instead of taking a rating at face value, you start by assuming a new place is probably about average. Then each review nudges it away from average.
A place with four reviews barely gets to nudge. A place with four thousand reviews gets to move the number wherever it wants. The more people who've weighed in, the more the rating counts for.
Say you're comparing restaurants in an area where the typical place has a 4.2 rating and about 400 reviews. Watch what happens when you weight things this way:
| Restaurant | Rating | Reviews | Adjusted |
|---|---|---|---|
| The brand-new place | 5.0 | 3 | 4.21 |
| The neighborhood favorite | 4.7 | 3,000 | 4.64 |
| The chain down the block | 4.1 | 1,100 | 4.13 |
The 5.0 collapses down to 4.21, because three reviews just aren't enough to prove anything. The 4.7 barely moves — three thousand people can't all be wrong. And the order flips to match what you'd have picked anyway if a friend had described all three places to you.
That's exactly what the Recommended sort does on MyTopRate. It runs this adjustment on every search, comparing each place against the other businesses that came back in your results.
Flipping to "most reviewed" has its own problem: you'll end up at the airport food court. Those places rack up thousands of reviews because thousands of people pass through, not because anyone likes them.
Review count tells you whether to trust a rating. It doesn't tell you the rating is any good. You need both numbers, read together.
Google shows you a breakdown of how many 5s, 4s, 3s, 2s and 1s a place got. Two restaurants can both average 4.3 for very different reasons.
One might be mostly 4s and 5s with a few grumps. That's consistently good. The other might be a big pile of 5s and a big pile of 1s that average out in the middle. That's a coin flip, and your dinner is the coin. A surprising chunk of 1-star reviews under a decent average is the most useful warning sign you'll find, and the average completely hides it.
Chefs quit. Restaurants get sold. A 4.8 built up over ten years can hide the fact that the last six months have been rough. Sort by newest, read five, and you'll know quickly whether you're looking at the place as it is now or as it used to be.
Bad reviews split into two kinds: stuff that would bother you, and stuff that wouldn't.
"Service was slow" is a dealbreaker on a 45-minute lunch break and completely irrelevant on a Saturday night with friends. "Too loud" matters if you're trying to talk and doesn't matter at all if you're celebrating. A 4.3 where all the complaints are about things you don't care about beats a 4.6 where the one recurring gripe is the exact thing you do.
That's it. Takes a minute and a half, and it beats trusting whatever sits at the top of the list.