How many ratings until a score is reliable?

The short answer: about 20

For each post we took the ratings in the order they arrived and compared the average after the first few with the post's average today. After one rating, the number was off by 1.57 points on average, and for 30.3% of posts it was off by 2 points or more. After five ratings the typical miss was still 0.84. After twenty it was 0.35, and 98.9% of posts were within one point of where they ended up.

So if you want one number to remember: under 10 ratings, a score is a rumor. Around 20, it is a fair estimate, good to within about a third of a point. Past 30, more ratings change little.

How far off the early average is, step by step

The table reads left to right. "Typical miss" is the average distance between the early score and the final one, in points on the 1 to 10 scale. The next columns show how many posts were already close, and how many were badly off.

Notice how slowly the first column shrinks. Going from 5 ratings to 20 means four times as many opinions, and the miss falls from 0.84 to 0.35. Going from 20 to 30 only takes it to 0.23. Each new rating helps less than the one before, which is why waiting for a few dozen is worth it and waiting for hundreds mostly isn't.

Why the first ratings swing so much

Two things make early averages jumpy. The first is that people simply disagree about the same photo. On posts with 10 or more ratings, the ratings spread about 2.1 points either side of the post's average, and 34.7% of posts received both a 1 and a 10 from different people.

The second is that raters have habits. Among people who have given 20 or more ratings, the strictest tenth average 3.77 or lower across everything they rate, and the most generous tenth average 7.44 or higher. If your first three raters happen to come from one end, your early score says more about them than about your photo. With twenty raters the strict and the generous mostly cancel out.

There is also a tilt. Early ratings run high: the first 3 ratings of a post average 0.54 points above its final score, and the first 10 are still 0.38 above. We can't say exactly why. The first people to see a new post are not a random sample of everyone who will see it later. Whatever the cause, a strong start that drifts down a little is the normal pattern, and it doesn't mean people turned against the photo.

How long it takes to collect enough ratings

Time is the practical cost of a reliable score. We looked at the 157 posts published since RATE's launch on September 27, 2026 that are at least a week old. The first rating usually arrives fast: the median wait is 1.8 hours. The tenth takes much longer. 140 of the 157 posts have reached 10 ratings, and for those the tenth arrived a median 59.9 hours after posting, about two and a half days.

Only 15.9% of posts had 10 ratings within a day, and 54.8% had them within three days. The twentieth rating arrived a median 140.2 hours after posting, almost six days. RATE is a small community, and these waits should shorten as it grows. For now, plan on about a week before you treat a score as settled. Rating other people's photos while you wait helps your own get seen more.

What this means when you compare two photos

Most people use ratings to choose between photos, and that is where small samples do the most damage. If each of two posts can be off by a third of a point at 20 ratings, a gap of 0.3 between them is noise. A few rules follow directly from the table:

• Wait until both posts have at least 20 ratings before you compare them. At 5 ratings each, about one post in thirteen is still off by 2 points or more.
• Treat a gap under half a point as a tie. Pick whichever photo you like more.
• A gap of a full point at 20 or more ratings is real. After 20 ratings, 98.9% of posts were within one point of their final score.
• Don't react to the first number you see. A new post usually opens a little high and comes down.
• One low rating is information about one person. In this data 34.7% of posts got both a 1 and a 10.

Why RATE shows no score at first

This is the reason a new post on RATE has no public score. Until enough people have rated it, the app shows that results are still being gathered. Showing the average after the very first rating would mean showing a number that is wrong by two points or more for nearly a third of posts, and people would make decisions and feel things based on it.

The exact number of ratings needed changes with how many people are active, so we don't print it here. How the score itself is put together is described on the page about how a RATE score is calculated.

Method and limits

We used the 89 posts with 40 or more counted ratings, as of November 11, 2026. Ratings were ordered by the time they were given, and the early score is the plain average of the first ratings in that order. A rating is counted only if it counts in the app; test, banned and deleted accounts are left out.

• The "final" score includes the early ratings, which flatters the later rows. Against only the ratings that came afterwards, the typical miss is 0.91 after 5 ratings, 0.73 after 10, 0.54 after 20 and 0.46 after 30.
• A final score from 40-odd ratings is itself an estimate. A post with 400 ratings might land somewhere slightly different.
• The same check on the 202 posts with 30 or more ratings gives a similar picture: a typical miss of 0.77 after 5 ratings, 0.55 after 10 and 0.28 after 20.
• This is one small community. On a 1 to 10 scale with a different crowd, the exact figures would differ, though the shape (fast improvement, then slow) is how averages behave everywhere.