How we score

Every listing on SiteList is crawled end to end and examined across 34 published dimensions. The score is computed, not negotiated — here is exactly how, including what we can’t measure.

The crawl

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 34 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	4
02	Audience & messaging	4
03	Usability	5
04	Accessibility	4
05	Design execution	5
06	Brand mark system	2
07	Performance	6
08	Imagery & art direction	2
09	Writing quality	4
10	Vertical credibility	3
11	Competitive position	3
12	Decision-support surfaces	2
13	Review-content integrity	2
14	Authority & link risk	2
15	Off-page readiness	2
16	Rank readiness	2
17	Risk & stability	3
18	Content briefs discipline	2
19	Editorial QA of content	3
20	Content program	3
21	Distribution & reach	2
22	Content freshness	2
23	Docs & self-serve help	3
24	Measurement readiness	2
25	Technical SEO	5
26	On-page SEO	4
27	Keyword targeting	3
28	Content portfolio health	3
29	Content gaps	2
30	Keyword gaps	2
31	Programmatic SEO quality	2
32	AI search readiness	4
33	Fix-priority hygiene	2
34	SEO composite coherence	1

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. Low-confidence assessments count at half weight. If more than 6 dimensions are missing, the review publishes with a visible “partial examination” banner.

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 AI review score (0–100) is the weighted mean of the 34 dimensions above — machine-examined evidence, recomputed on every crawl.

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 says “Unverified by users,” and we never emit a placeholder star rating anywhere, including structured data.

What money can and can’t buy

Owners can pay for tools: re-crawls, 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 crawler fails at something, the review’s methodology footer lists the gap instead of papering over it.

Questions about a specific verdict? Find the review — every claim in it cites the evidence it stands on.