How We Score Products
Last updated: October 7, 2026
Every product we review gets a FindorLabs Score out of 10 and an evidence level (High, Medium or Low). The score tells you how good we think the product is. The evidence level tells you how much proof stands behind that score.
The formula
FindorLabs Score = 30% Build + 40% Performance + 30% Value
| Sub-score | The question it answers | What we look at |
|---|---|---|
| Build (30%) | Will it last? | Durability and reliability mentions on Amazon, failures described in the critical reviews we read, recalls, warranty |
| Performance (40%) | Does it do its job well? | Independent lab or tester results, when they exist, and the share of positive performance mentions from buyers |
| Value (30%) | Is it worth the price? | Value-for-money mentions, and what you get compared with similar models at that price tier |
Performance weighs most because it’s why you buy the product. Each sub-score goes from 0 to 10 in half-point steps, and the total is rounded to one decimal.
Where the evidence comes from
- Buyer reviews from several retailers (Amazon, Walmart, Target, Best Buy), with a focus on the critical ones (one to three stars), because they show what goes wrong.
- Amazon’s own counts of what reviewers mention (for example, “noise: 574 of 699 mentions negative”). We don’t use Amazon’s AI-generated summaries.
- Independent labs and testers such as RTINGS, when they have tested the exact model. We say when a source earns affiliate commissions.
- Official sources: manufacturer specs (labeled as claims), warranty terms and U.S. Consumer Product Safety Commission recalls.
Evidence level: High, Medium or Low
| Level | What it means |
|---|---|
| High | Enough critical reviews read, at least three independent sources, and an independent test of this exact model |
| Medium | Enough reviews read, but no independent test of this exact model, or most of the evidence comes from one retailer or is copied from the brand’s own site |
| Low | Too few reviews or a single source. We don’t publish a score at this level; we say the data is too thin |
A great product can have Medium evidence (it’s new and nobody has tested it yet), and a mediocre one can have High evidence. The two answer different questions.
What we don’t do
- We don’t claim tests we didn’t run. Our reviews are research-based: we analyze buyer reviews and published tests. If we ever test a product ourselves, the review will say exactly what we tested and carry a Hands-On badge.
- Brands can’t buy a score. We earn affiliate commissions when you buy through our links, at no extra cost to you, and that never moves a number.
- We don’t hide disagreement. When two labs or buyers and labs disagree, we tell you.
A worked example: the Ninja Crispi (6.4/10)
From our Ninja Crispi review, September 2026:
| Sub-score | Evidence | Score |
|---|---|---|
| Build | 78% of Amazon durability mentions negative; “LID” error failures reported on Amazon, Target and Best Buy; glass breakage rare; 1-year warranty | 5.5 |
| Performance | 90-94% of Amazon cooking and crisping mentions positive; RTINGS rates its frying mid-range; no temperature control | 7.0 |
| Value | 66% of Amazon value mentions positive; the Crispi Pro fixes its main limits for more money | 6.5 |
Calculation: 0.3 × 5.5 + 0.4 × 7.0 + 0.3 × 6.5 = 1.65 + 2.80 + 1.95 = 6.4
Evidence level: High. We read 119 low-star Target reviews and about 30 Amazon critical reviews, used four retailers plus two independent testers who tested this model (RTINGS and Prudent Reviews), and weighted down reviews copied from the brand’s sites.
The details, for the curious
How sub-scores are set. Each sub-score is an editorial judgment made with the same reference points for every product:
| Range | Build | Performance | Value |
|---|---|---|---|
| 8.0-10 | Durability mentions mostly positive (about 75% or more), failures rare in critical reviews, no recall | Top or near-top result in independent tests, and strong buyer praise | Value mentions mostly positive, and as good or cheaper than similar models |
| 6.0-7.5 | Mixed durability, or one recurring defect in a minority of critical reviews | Good test results or strong buyer praise, with a known limitation | Fair price for what you get |
| 4.0-5.5 | Most durability mentions negative, or failures in roughly a quarter or more of the critical reviews we read | Clear weakness on a core task | Price not backed by build or performance, or value mentions mostly negative |
| Below 4 | Safety recall or failures dominating the reviews | Fails at its main job | Not worth buying at any price |
How we handle tricky reviews.
– Copied reviews: many retailers show reviews copied from the brand’s own website. We count them once and give them less weight.
– Incentivized reviews: reviews marked as collected in a promotion or tester program get less weight.
– Duplicates and wrong products: reviews attached to the wrong listing are excluded.
– Critical reviews are a biased sample: they show what goes wrong, not how often. We never present them as failure rates.
When we don’t score. If a product has too little independent data (for example, fewer than about 100 topic mentions on Amazon and no independent test), we mention it without a score.
Updates. Ratings and review counts carry the date we checked them. We recheck our picks when new models launch or the evidence changes, and the review shows its last update date.
Questions about a score? Contact us.