Methodology
The long version of how ReviewCatalog scores products — how the rubric is weighted, how we read buyer evidence, how we handle conflicts of interest, and when we walk away from a review entirely.
This page exists because the shorter How We Review page glosses over the trade-offs. The methodology below is what we actually do, including the parts that are messy or imperfect.
1. What the verdicts are built from
We don’t run a lab, and we don’t pretend to. Every verdict is built from the largest body of evidence that exists for a marketplace product: the buyers. For each product we pull together hundreds of real AliExpress reviews across the top listings — sorted by most recent and by lowest rating, where the honest reporting lives — plus the listing’s own claims, specs, price history, and the seller’s track record. The skill is in the reading: one angry buyer is noise; the same complaint from a dozen buyers across different sellers is the truth about the product.
2. The 100-point rubric
Every product is graded on the same five categories. Each maps to a specific way AliExpress purchases go wrong:
- Build quality (25 pts) — what buyers consistently report about materials and durability: the lamp that falls apart in a week, the strap that won’t stay closed.
- Performance (30 pts) — whether it does what the listing says, judged on how reliably buyers confirm the claims. Inflated capacity and brightness numbers die in this category.
- Value for price (20 pts) — graded against the price actually charged, including shipping. A $9 item is judged by $9 standards.
- Match to description (15 pts) — how often buyers report the product differs from the photos, dimensions, or materials in the listing.
- Safety and packaging (10 pts) — reports of transit damage, missing parts, or hazards, especially on anything electrical.
Performance carries the heaviest weight because that’s where the biggest dollar mistakes happen. Safety has the smallest numerical weight only because it works as a kill switch: a credible safety pattern in the reviews drops the product to “do not buy” regardless of its other scores.
3. How products get picked
A product earns a review when all four of these line up:
- Demand — it (or close variants) sells in real volume, so the question matters to enough people.
- Review density — at least 200 buyer reviews exist across the top listings, so there’s a pattern to read rather than anecdotes.
- Listing confusion — the top listings include near-identical products at very different prices. If a buyer can pick blindly and be fine, we’re not needed.
- We can judge it from evidence — no medical, prescription, or safety-critical products whose claims only a lab can verify. We’d rather skip a category than fake authority in it.
4. Fake and incentivized reviews
Buyer reviews are the raw material, so gaming them is the main threat to the method. We discount walls of five-star one-liners posted close together, reviews that describe a different product (a reused listing), and ratings that contradict their own text. We weight recent, critical, photo-backed reviews the heaviest — they’re the hardest to fake at scale. The same checks we run are described for readers in our AliExpress safety guide.
5. Updates and corrections
AliExpress prices and stock move constantly. If the price moves more than 20% from what we recorded, or the dominant seller is replaced, we revise the review within 30 days; the “updated” date sits at the top of every review. When readers report an experience that contradicts a verdict, we re-read the current review pattern — products change batches, and a good product can go bad. Corrections get made in the text, not hidden.
6. Conflicts of interest
Revenue comes from AliExpress affiliate commissions. The commission is the same whether a verdict is “buy it” or “skip it”, which removes the incentive to inflate scores — and in practice roughly a third of covered products end with a “don’t bother.” We don’t take payment for placement, we don’t accept sponsored verdicts, and no seller sees a review before it’s published. The full disclosure is here.
7. What this method can and can’t tell you
Read honestly, aggregated buyer evidence is very good at three things: whether a product matches its listing, how it holds up over the first weeks and months across many households, and which specific flaw it tends to have. It cannot measure what only instruments can — exact capacity, emission spectra, material composition. Where a claim can only be verified in a lab, we say so in the review instead of pretending. That line is the whole difference between a review site you can trust and one that tells you what you hoped to hear.
Questions about the method? hello@reviewcatalog.com — or see how the desk works.