How we score
Every listing on SiteList is reviewed end to end across 13 public dimensions. The score is computed, not negotiated — here is exactly how, including what we can’t measure.
What we check
We fetch up to 40 pages per site: the homepage, navigation pages, pricing, docs, blog samples, legal pages, plus probe requests (broken-URL handling, redirect behavior, feeds, llms.txt). We capture rendered HTML, full-page screenshots on desktop and mobile, network logs, CSS, structured data, and run Google PageSpeed measurements. Public records fill in the rest: archive history, domain registration, page-rank estimates, entity lookups.
Everything a review claims traces back to one of those artifacts. If we couldn’t fetch something, the review says so in its methodology footer.
The 13 public dimensions
Each dimension is scored 0–100 by a specialist analysis module with a fixed, published weight. The weights sum to exactly 100 — no dimension is secretly more equal than others.
| # | Dimension | Weight |
|---|---|---|
| 01 | First impressions & positioning | 8 |
| 02 | Audience & messaging | 8 |
| 03 | Usability | 10 |
| 04 | Accessibility | 8 |
| 05 | Design execution | 10 |
| 07 | Performance | 12 |
| 09 | Writing quality | 8 |
| 12 | Decision-support surfaces | 4 |
| 13 | Review-content integrity | 4 |
| 17 | Risk & stability | 6 |
| 19 | Editorial QA of content | 6 |
| 23 | Docs & self-serve help | 6 |
| 25 | Technical SEO | 10 |
When a dimension doesn’t apply (a product with no docs subtree, say), it is excluded and its weight is redistributed pro-rata — never silently zeroed. This is the v2 public score set; former 34-dimension scores are archived when a listing is migrated. Low-confidence assessments count at half weight. A new review publishes only after every applicable public dimension has a valid score. If a public dimension fails or is absent, the review is held for a stage retry or an administrator’s QA decision; we do not publish an incomplete score by redistributing around the failure. The “partial review” banner remains on older published reviews and reviews built from truncated crawl coverage, where it identifies the limits of the available evidence.
Score bands
- 0–39Critical issues
- 40–59Below the bar
- 60–74Fair
- 75–89Strong
- 90–100Exceptional
Bands are discrete: a 74 and a 75 sit in different bands, on purpose. A continuous gradient would imply precision the method doesn’t have.
Two scores, never blended
The SiteList review score (0–100) is the v2 weighted mean of the 13 public dimensions above — measured automatically and recomputed on every review. Paid fresh reviews add private diagnostics and strategy without changing this public score.
The SiteList score (0–5 stars) comes from verified user reviews: 40% credibility-weighted user rating, 30% feature completeness against the category baseline, 20% review depth, 10% recency. It renders only once a product has at least 3 approved user reviews — below that a listing simply shows no star rating, and we never emit a placeholder star rating anywhere, including structured data.
What money can and can’t buy
Owners can pay for tools: private diagnostics from fresh reviews, widgets, marketplace offers, analytics. Owners cannot pay to change a score, reorder a category, or remove an accurate finding. Incentivized user reviews always carry an FTC disclosure we write and they cannot edit. Comparison “winner” cells are computed from facts with cited sources.
Reviews are generated by software and say so — the byline is “SiteList Engine,” never a fictional person. When our automated review cannot assess something, the review’s methodology footer lists the gap instead of papering over it.
Questions about a specific review? Find the review — every claim in it cites what it is based on.