Methodology

How national offer scores work

The main RealReviews score is our determination of user consensus across eligible, inspectable sources. It is not a rescaled rating or an average of outside platform stars.

Consensus before arithmetic

Researchers search for all eligible user-experience evidence, inspect actual content-bearing reviews and complaints, and code recurring positives, recurring negatives, disagreements and counterexamples. A source is eligible only when the entity match is defensible, the experience is attributable enough to analyze, bodies are inspectable and the collection context can be disclosed. Thin question threads, broker promotion, wrong-entity results and rating-only pages are documented but excluded from scoring.

What enters the score

The score uses only coded user experience. Dimensions vary when a category genuinely requires different decision criteria, but the record must show the weights, inspected-body count, eligible sources, confidence, product specificity and written rationale. For merchant cash advances, the six dimensions are access and speed, application service, terms clarity, affordability and payment burden, servicing and workouts, and repeat use and outcomes.

What does not enter the score

Every platform's displayed rating, review count, capture date and collection context remain separately attributed. RealReviews never averages those stars or rescales one platform's rating. Official product pages, contracts, company records, laws and regulator sources verify identity and supply the context needed to interpret a user report; they earn no score points. Affiliate status also carries zero weight.

What the 0–10 scale means

RangeEditorial meaning
0.0–2.9Overwhelmingly adverse consensus with serious recurring experience problems.
3.0–4.9More negative than positive consensus, including material recurring weaknesses.
5.0–6.9Mixed consensus: meaningful positives coexist with substantial or consequential negatives.
7.0–8.4Strong positive consensus with material limitations or credible counterexamples retained.
8.5–10Exceptional positive consensus across eligible evidence; never a guarantee of an individual outcome.

Scores can change

Review evidence and collection contexts can change. Every score is dated. A meaningful change requires a new source search, body audit and score record rather than silently overwriting the determination. When the evidence is too thin or too ambiguous to support consensus, the public result should be Not scored rather than a number inferred from product terms alone.

Affiliate relationships do not enter the formula

Commission rate, affiliate approval, sponsorship, and advertiser status carry zero weight. RealReviews may earn a commission from a recommended or poorly scored offer, but the relationship must be disclosed and cannot alter criticism, alternatives, ranking order, or the score.