Scoring Standards — How We Rate Products
Our Scoring Methodology
We do not physically test products. Instead, we analyze 67M+ verified Amazon customer reviews to derive data-driven quality scores. Every score is calculated using a transparent, publicly documented formula.
The ReviewScore Formula
Each product receives a ReviewScore from 0-100, calculated across four core dimensions:
ReviewScore = w1 × ComplaintScore + w2 × AuthenticPraiseScore + w3 × DurabilityScore + w4 × RecommendScore
| Dimension | What It Measures | Data Source |
|---|---|---|
| ComplaintScore | Absence of specific, verifiable defects (e.g., “coating peeled”, “stopped working”). Higher = fewer complaints. | 1-2★ negative reviews |
| AuthenticPraiseScore | Presence of specific, detailed praise (e.g., “even heating across the surface”). Generic “great product” reviews are excluded. | 4-5★ positive reviews, AI-filtered |
| DurabilityScore | Evidence of long-term reliability — reviews mentioning use after months/years with continued satisfaction. | Reviews with time markers (“after 6 months”, “still working after 1 year”) |
| RecommendScore | Explicit willingness to repurchase or recommend (“would buy again”, “highly recommend”). | Review text pattern matching + AI semantic analysis |
Category-Specific Weights
The weights (w1-w4) vary by product category because what matters for cookware differs from what matters for curtains:
- Cookware: Durability and complaint rate dominate (w1=0.30, w3=0.30) — coating lifespan is critical.
- Curtains: Appearance and satisfaction matter more than durability (w2=0.35, w4=0.30).
- Vacuum Cleaners: Durability is paramount (w3=0.30) — motor/battery lifespan is a top concern.
See each category’s scoring dimension page for exact weights and category-specific bonus/penalty keywords.
Bonus & Penalty Keywords
Each category has bonus keywords (features users actively seek) and penalty keywords (pain points users complain about). When these keywords appear in reviews, they adjust the score:
- Bonus: +2 to +5 points when reviews mention specific desired features (e.g., “even heating” for cookware).
- Penalty: -3 to -8 points when reviews mention specific defects (e.g., “coating peeled” for cookware).
Bonus/penalty keywords are derived from our pain point discovery process using AI analysis of thousands of negative reviews per category.
AI Review Authenticity Filter
Not all reviews are useful. Many are generic (“great product”, “love it”, “very good”) and provide no actionable insight.
We use AI to classify every sampled review as:
- Specific: Mentions concrete features, use cases, comparisons, materials, or detailed experiences. → Included in scoring.
- Generic: Only expresses vague sentiment without detail. → Excluded from scoring.
This ensures our scores reflect real, verifiable user experiences — not empty praise.
Review Confidence
Every product score is accompanied by a Review Confidence indicator based on the number of reviews analyzed:
| Reviews Analyzed | Confidence Level |
|---|---|
| 1,000+ | ★★★★★ Very High |
| 500-999 | ★★★★ High |
| 100-499 | ★★★ Medium |
| <100 | ★★ Low (use with caution) |
Limitations
- Scores are based on review data analysis, not physical product testing.
- Review samples may not represent all purchasers equally.
- Scores are recalculated periodically as new reviews become available.
- Products with very few reviews may have unreliable scores.
Updates
Scoring weights and bonus/penalty keywords are reviewed periodically. Each category page shows its last update date. Scores for individual products are recalculated each time we refresh our review analysis.
⚠️ Affiliate Disclosure: As an Amazon Associate we earn from qualifying purchases. Product links on this site use Amazon affiliate tags. This does not affect our scoring — scores are calculated independently from our monetization.