Expert reviews. Community sentiment. Price history. Aggregate scoring. No paid placements, no astroturfed reviews, no brand relationships.
We aggregate numeric scores from published expert reviews. Tests use standardized measurement equipment — frequency response, distortion, isolation — not gut feelings. Each source is normalized to a 0-100 scale based on its original scoring system.
Reddit threads and posts are scraped and analyzed for sentiment. Extreme opinions (very positive or very negative) signal consensus — products that everyone loves or everyone warns against. Astroturfed marketing campaigns are filtered out by cross-referencing post timing against review volume spikes.
Prices are tracked daily. We flag products where the "original" price was raised recently to make a discount look larger. If the 30-day average price is significantly below the listed original, the discount is genuine. If the original was inflated last week, we remove the deal badge.
What each score range means in practice
Routinely recommended by experts. Genuine category leaders.
Solid products with verified expert approval. No major flaws.
Mixed reviews or niche appeal. Check pros/cons carefully.
Significant expert or community concerns. Usually discounted heavily for a reason.
Insufficient expert coverage. May not have been reviewed by credible sources.
How we arrive at every score you see
TweakDeals was built to solve a specific problem: AliExpress is a marketplace where quality varies wildly, discounts are often inflated, and it is nearly impossible to tell whether a product is genuinely good or just well-marketed. Our scoring pipeline was designed to cut through that noise by combining four independent layers of evaluation — each one cross-checking the others.
The foundation of every score is expert review aggregation. We pull numeric scores from established review sources (Crinacle/IEF, AudioScienceReview, Rtings.com, TechPowerUp, and others) that use standardized measurement equipment — frequency response graphs, distortion measurements, isolation tests — rather than subjective impressions. Each source is normalized to a 0-100 scale so that a product scored by multiple reviewers produces a single meaningful average. Older reviews are penalized to favor recent data, since product revisions and firmware updates can significantly change real-world performance.
To this expert baseline we add community sentiment analysis, drawn from Reddit discussions and community forums. This captures long-term reliability issues, firmware bugs, and customer support experiences that expert reviews — typically based on a single unit over a short period — may miss. Because community sentiment is more volatile than expert evaluation, it applies as a modifier of at most ±5 points, ensuring that a handful of vocal users cannot distort a product’s score. We also filter for coordinated marketing campaigns by cross-referencing post timing with review volume spikes.
Our price analysis layer tracks daily price data and flags products where the “original” price was recently raised to exaggerate a discount. If the 30-day average price is significantly below the listed original, the discount is genuine. If the original was inflated in the last week, the deal badge is stripped and the product receives a lower deal score. This prevents the common AliExpress practice of listing products at inflated MSRPs to make every sale look like a bargain.
The final score you see on each product page is the Aggregate Score, which combines the expert mean (with age penalties), the Reddit sentiment modifier (clamped at ±5), and the deal score weighted at 40%. A product can have a high quality score but a low deal score (if it is overpriced), or a modest quality score but an excellent deal score (if the discount is genuine and deep). Both numbers matter, and we surface them independently so you can make your own judgment. All scores are recalculated weekly; prices update daily. We do not accept payment from brands to alter scores, and our affiliate commissions are rate-identical regardless of score.
We track prices daily. If a product's "original" price was raised in the last 30 days specifically to make a discount look larger, we flag or remove it. Products where the AliExpress "discount" is inflated beyond realistic value receive a lower deal score.
Products without expert reviews or community discussion are listed without a score. We won't fabricate a score from insufficient data. If a product hasn't been reviewed by experts or discussed by real users, we show the price but no score.
Quality score = expert + community evaluation of the product itself. Deal score = value + price trajectory. A $10 product with a 60 quality score and a genuine 30% discount gets a higher deal score than a $200 product at 5% off.
Reddit community sentiment is analyzed separately from expert reviews and applies as a modifier of at most ±5 points to the aggregate score. This captures real-world long-term ownership issues (reliability, firmware bugs, customer support) that expert reviews don't always surface. Expert reviews carry more weight because they use measurement equipment and standardized testing.
No. We have no paid placements, no affiliate arrangements with brands, and no sponsored reviews. Our income comes from AliExpress affiliate links, which are the same rate for all products regardless of score.
Scores are recalculated weekly. Prices are tracked daily. A product's score only changes when new expert reviews or significant community discussion surfaces — not when price fluctuates.
Browse products with real scores, price history, and expert analysis.
