Almost every product on AliExpress is rated between 4.6 and 5.0. When nearly everything is excellent, the rating has stopped separating good from bad, and the buyer who relies on it is choosing at random with extra confidence. This guide to AliExpress ratings and reviews explains where each number comes from — star rating, review count, sold count, store feedback — why the ratings compress towards the top, and how to read them in combination so they mean something again. It also explains why the AliScore extension weights them the way it does.
The four numbers on a listing
Star rating
The mean of buyer ratings on that listing, from 1 to 5. Every listing has one once a few reviews exist. It carries the most weight in any buyer's head and the least information per digit, for reasons covered below.
Review count
How many rated orders sit behind the average. This is the number the star rating quietly depends on and the one the page makes easiest to miss. It is the single best indicator of how much to believe the rating at all.
Sold count
Orders placed, across variants, for the listing. It measures popularity, not satisfaction — a product can sell thousands and disappoint most of them. It is useful as a tie-breaker and as evidence that a listing is real, and it is what AliScore falls back to for sentiment when a listing has sales but no review statistics yet.
Store positive feedback rate
A store-level figure, not a product-level one: the share of feedback across the store's recent orders that was positive. It is compressed even harder than the star rating, with almost every store between 90% and 100%. Our seller ratings guide gives the strict scale for reading it.
Why everything is 4.7
Three forces push AliExpress ratings up and squash them together:
- Prompted ratings. The app asks buyers to rate. The path of least resistance is five stars with no text. A large share of ratings are this — satisfied enough not to complain, not engaged enough to describe.
- Survivorship in disputes. A buyer who receives nothing or receives the wrong item often opens a dispute rather than a review. The worst outcomes are under-represented in the average.
- Selection by the seller. Listings with poor ratings get relisted under a new item ID, resetting the average. The survivors are the ones that rated well.
The result is that a 4.6 and a 4.9 are nearly the same product, statistically. The rating still has value at the extremes — a listing that has managed a 4.3 despite all this has real problems — but in the 4.6 to 5.0 range where nearly everything sits, it is close to noise.
What separates products, then
If the average does not, three other things do.
1. The count behind it
Ten five-star ratings is a small sample subject to every bias above at full strength. A thousand ratings averaging 4.7 is a population, and the biases are baked in evenly across every competitor, so the comparison is fair. Below ten reviews, ignore the rating entirely. Between ten and a hundred, treat it as a rough indication. Above a few hundred, trust the comparison between listings.
This is why AliScore applies what it calls a confidence dampener: below ten reviews, the total score is scaled down, and a listing with a single review keeps roughly half of whatever it earned. Without that rule every brand-new listing with one glowing review would show as Excellent.
2. The negative share
The average hides the shape. A 4.7 can be mostly fives with a few threes, or a mix of fives and ones. The second is a product that works for most buyers and fails badly for some — a lottery — and the average will never tell you. AliExpress publishes the positive-versus-negative split per product; once negatives exceed roughly one review in ten, the listing has a pattern of failure, not just noise. AliScore reads that split directly and penalises the score past the 10% line.
3. Volume of evidence over time
A listing with reviews spread across two years has been bought, used and reviewed under real conditions, including after the honeymoon. One with all its reviews from the last month has not yet had a chance to fail.
How AliScore weights these
The extension's score is four components combined, each on a 0–10 scale, written out in full on the how it works page:
| Component | Weight | Built from |
|---|---|---|
| Quality | 35% | The star rating, normalised to 0–10 |
| Sentiment | 25% | Positive share of reviews, penalised past 10% negative, nudged up by a strong "Quality" label |
| Review volume | 25% | A logarithmic curve over the review count |
| Value | 15% | Rating against price on a log scale, plus credit for a genuine discount |
Then the confidence dampener applies below ten reviews. The star rating still carries the most weight, because it is the one signal every listing has — but half the score is about how much evidence there is and what the negatives say, which is exactly the information the listing page buries.
The score is cut into plain verdicts on the pill itself: 8.0 and up is Safe to buy, 6.0 to 7.9 is Worth considering, 4.0 to 5.9 is Buy with care, and below 4.0 is Risky. A listing with no data gets no verdict at all — "not enough data" rather than an invented number.
Reading a listing in thirty seconds
- Find the review count. Under ten: unrated, whatever the stars say.
- Find the negative share, or read the one- and two-star reviews. Over one in ten: a pattern.
- Check the date spread. All recent: no long-term evidence.
- Only then read the star rating, and compare it to competitors with similar counts.
- Open the store and read its feedback rate on the strict scale.
Or install AliScore and read one number with its breakdown open. Either way, the rating stops being the first thing you look at, which on AliExpress is the whole trick. For judging whether the reviews behind the numbers are genuine, see how to spot fake reviews.
Spotted a mistake or something out of date? Mail anwarazim03@gmail.com with the URL — one person maintains this and reads every message.