Loading…
Loading…
A score is only as trustworthy as the method behind it. Here, in full, is how we get from a web address to a verdict — so you can judge the tool, not just the number.
Whole-website assessments start with the site's sitemap.xml, falling back to a link crawl, and respect robots.txt. Crawling is bounded by a depth and a page cap, so very large sites may be reported as covering a subset of pages.
Each page is assessed with the Ascent Accessibility engine — an in-house, clean-room, open-source WCAG rule engine — against the rules for the standard you selected. Automated rules detect only issues that can be identified automatically, roughly 30–50% of WCAG criteria. A companion site audit measures performance, SEO, and best-practice signals, and an optional AI-assisted review narrows the items that require manual judgement.
The score starts at 100 and subtracts a fixed weight per finding — critical −10, serious −5, moderate −2, minor −0.5 — capped per finding. 90+ is a pass, 70–89 is partial, and below 70 is a fail. The conformance table separately tracks which success criteria pass, fail, are not applicable, or need manual review.
Automated tools cannot judge meaning, so many criteria (for example, whether alt text is accurate, or whether content is genuinely understandable) require human review. A clean score is encouraging, but it is not a certificate of conformance. Treat the report as a starting point for a manual review.
Criteria the engine and AI cannot determine are resolved by a certified reviewer with lived experience of the barrier they affect. Every reviewer is certified through the Ascent Web Accessibility Program.
| Criterion | Reviewed by | Why human review |
|---|---|---|
| 1.2.8 Media Alternative (Prerecorded) | a certified reviewer with lived experience of deafness (captions/media) | Only a person can confirm a video's full text alternative conveys the same information. |
| 1.4.8 Visual Presentation | a certified reviewer with lived experience of low vision | Whether custom presentation settings truly help needs a low-vision user's judgement. |
| 2.1.3 Keyboard (No Exception) | a certified reviewer with lived experience of limited movement (keyboard-only) | Only a keyboard-only user can confirm every function is genuinely operable. |
| 2.1.4 Character Key Shortcuts | a certified reviewer with lived experience of speech/voice control | Only a voice-input user can tell whether character-key shortcuts interfere. |
| 2.2.4 Interruptions | a certified reviewer with lived experience of cognitive or learning disability | Whether an interruption is tolerable needs a person's judgement. |
| 2.2.5 Re-authenticating | a certified reviewer with lived experience of cognitive or learning disability | Whether re-authentication preserves data needs a person to verify the flow. |
| 2.2.6 Timeouts | a certified reviewer with lived experience of cognitive or learning disability | Whether a timeout warning is adequate needs a person to test it in context. |
| 2.3.1 Three Flashes or Below Threshold | a certified reviewer with lived experience of photosensitive epilepsy | Whether flashing content is safe requires a person to review it. |
| 2.3.2 Three Flashes | a certified reviewer with lived experience of photosensitive epilepsy | Whether content ever flashes above the threshold requires a person to review it. |
| 2.5.6 Concurrent Input Mechanisms | a certified reviewer with lived experience of limited movement (keyboard-only) | Whether concurrent input mechanisms work needs a person to test them. |
| 3.1.6 Pronunciation | a certified reviewer with lived experience of cognitive or learning disability | Whether pronunciation guidance is adequate needs a reader's judgement. |
| 3.2.5 Change on Request | a certified reviewer with lived experience of cognitive or learning disability | Whether context changes are user-initiated needs a person to test the flow. |
| 3.3.4 Error Prevention (Legal, Financial, Data) | a certified reviewer with lived experience of cognitive or learning disability | Whether error prevention for legal/financial submissions works needs a person to test it. |
| 3.3.6 Error Prevention (All) | a certified reviewer with lived experience of cognitive or learning disability | Whether error prevention for all submissions works needs a person to test it. |
| 3.3.8 Accessible Authentication (Minimum) | a certified reviewer with lived experience of cognitive or learning disability | Whether authentication avoids cognitive tests needs a person to attempt the flow. |
| 3.3.9 Accessible Authentication (Enhanced) | a certified reviewer with lived experience of cognitive or learning disability | Whether authentication avoids cognitive tests entirely needs a person to attempt the flow. |
| 1.2.4 Captions (Live) | a certified reviewer with lived experience of deafness (captions/media) | Live captions must be reviewed by a person who relies on them. |
| 1.2.9 Audio-only (Live) | a certified reviewer with lived experience of deafness (captions/media) | A live audio alternative must be reviewed by a person who relies on it. |
Our engine is clean-room and open source — built from the W3C's own success criteria and the ACT-rules format, not from any third-party scanner. That means every rule can be inspected, and the engine benchmarked against the W3C/WAI test suites. We publish the method precisely so the evidence behind your report is verifiable.
See the full list of WCAG 2.2 success criteria the tool scores against.