Recommendations Matter

How the score works

Every score on this site is published together with the sample it stands on. Here is the formula, in full.

The formula

Every Reddit mention of a product is classified as positive, negative or neutral. Neutral mentions — questions, bare specs, passing references — are counted but excluded from the score: a mention with no stance is not evidence of satisfaction.

n     = positive + negative mentions        (the opinionated sample)
p     = positive share of n, weighted by engagement, specificity and topic
score = Wilson lower bound (95%) of p over n, ×100

The Wilson lower bound is what makes the score volume-aware: a product needs both a high positive share and enough opinions to score highly, so five glowing comments cannot outrank two hundred mostly-positive ones.

A product with no opinionated mentions has no score at all — we show its mention count instead. Internally such a product is floored at the bottom of the ranking, so an unevidenced product can never occupy a top slot.

What each number means

Mentions

Every relevant Reddit post or comment we matched to the product — positive, negative and neutral together. This is a measure of how much the product is discussed, not of how well it is liked.

n — the opinionated sample

The positive and negative mentions only. This is the number the score is computed over, and it is the number to look at when you ask how much a score is worth. n is always smaller than mentions whenever neutral discussion exists.

% positive

The positive share of n, after weighting. Each opinion carries a weight of topic × (1 + specificity) × (1 + engagement), so a detailed comment in an on-topic thread with real upvotes counts for more than a bare “this one’s good”.

Score

The 95% Wilson lower bound of that positive share over n, on a 1–100 scale. Read it as: “given this sample, we are 95% confident the true positive rate is at least this”.

Why a lower bound instead of a plain percentage

A raw percentage ranks 5 positive opinions out of 5 (100%) above 190 out of 200 (95%), which gets it backwards: the first result can collapse with a single further opinion, the second cannot. The Wilson lower bound discounts a share by how little evidence stands behind it, so a five-opinion sample is pulled far below its face value while a two-hundred-opinion sample barely moves.

The practical consequence: a product can lose to a rival with a lower positive share simply because far more people weighed in on the rival. That is intended.

What this score does not tell you

It measures what Reddit says, not how a product performs on a bench. We do not test products.
Sentiment is assigned automatically, so individual mentions can be misread. The larger n is, the less any single misread matters.
A niche product can be excellent and still score low simply because few people discuss it. Check n before treating a low score as a verdict.
Scores move as new discussions are picked up — they are a snapshot, not a permanent grade.

See it applied

Every product card shows its mention count and its opinionated sample n next to the score, so you can judge the number without taking our word for it.

Browse products About us