THE METHODOLOGY
Know what
the numbers mean.
Our assessment is a transparent view of available technical evidence. It is a starting point for improvements.
Six categories. Product-defined weights.
| Category | Weight | Scoring |
|---|---|---|
| Performance | 25% | Mean of available mobile and desktop Lighthouse performance scores. |
| Search readiness | 20% | Title and crawl directives weight 2 each; description, canonical, H1, alt attributes, Open Graph, robots, sitemap and limited links weight 1 each. |
| Enquiry pathways | 25% | Phone link weight 2; enquiry pathway indicators weight 3; action wording weight 1; manual delivery review weight 1 excluded from score. |
| Mobile & accessibility | 15% | Viewport weight 2; zoom configuration weight 1; available Lighthouse checks weight 1 each; manual visual review weight 1 excluded. |
| Security configuration | 10% | Verified HTTPS weight 3; mixed references, four header-presence checks and HTTP upgrade weight 1 each. |
| Local & AI readiness | 5% | Typed JSON-LD weight 2; identity and semantic indicators weight 1 each; local search manual review weight 1 excluded. |
Missing is not zero.
A category score is the weighted proportion of passed checks among available pass/fail checks. At least half of that category’s defined check weight must be observable to calculate a category score. Manual and unavailable checks are excluded from the score. In performance, one available device test is enough, with 50% category completeness.
The composite is the weighted average of available category scores, renormalised over their category weights. Completeness is the sum of each category’s available check proportion multiplied by its product weight. Manual reviews reduce completeness and are never silently marked as passed. A high score with incomplete evidence is not a clean bill of health.
What we actually inspect.
Homepage HTML and response headers, the HTTP upgrade path, robots.txt, one same-origin sitemap candidate, and up to three selected same-origin links. The HTML inspector does not execute JavaScript, assess visual prominence, validate every structured-data type, or submit forms. HTTPS trust and hostname checks are performed on the actual connection. Missing headers indicate configuration, not compromise.
Lighthouse uses Google’s remote browser. Its laboratory metrics are labelled separately from available field measurements. INP is shown only when supplied in field data, with page or origin scope. TBT is a lab metric and is never relabelled INP. Field CLS percentiles are converted from the provider’s scaled representation.
Signals have limits.
CTA, booking, enquiry and business-identity detection uses HTML indicators. A feature may exist in JavaScript or on another page. A missing indicator is a prompt for review. No ranking, traffic, lost-customer or revenue estimate is inferred. AI readiness is a signal assessment, not a ranking score.
Prospect priority is separate.
The private opportunity score awards up to 30 points for verified failed checks (5 each), 15 for conversion indicators (5 each), 15 for evidence completeness, 20 for manually assessed business fit and 20 for manual qualification. An administrator can override the score. A weak website alone is not proof of a valuable opportunity.
Cached results and priorities.
Public checks may reuse technical findings for the same full URL and visibility for up to six hours. The original measurement time remains visible. Private prospect reruns bypass this cache. Priorities select up to three failed checks, ordered by severity and then verified evidence. Suggestions and manual checks are shown separately.
Check your website