Methodology

A score you can explain.

Stepfree uses transparent automated checks to detect regressions. It keeps machine findings and human judgement deliberately separate.

What we measure

Our scanner evaluates the live page after JavaScript has run. It checks computable WCAG patterns such as missing accessible names, contrast thresholds, document language, form labels, and heading structure.

Automated tools can detect roughly 25-35% of WCAG issues. Whether alternative text is meaningful, focus order is logical, or an error message is helpful still requires a person, often using assistive technology.

Stepfree ScoreAutomated checks

Each confirmed finding is weighted by its axe-core impact. Higher-impact findings affect the score more. The result is normalised by pages scanned and clamped from 0 to 100.

10

Critical

7

Serious

3

Moderate

1

Minor

Published formula

100 × (1 − weighted findings ÷ (pages scanned × 12))

The normalisation constant is 12. Human-review items are never counted as confirmed findings.

How results are handled

Confirmed findings

Rules with enough evidence to report as conformance issues detected. These contribute to the score.

Needs human review

Cases where automation cannot decide. These stay separate and never count as failures.

New and resolved

Stable fingerprints compare each run with the previous scan so your team can focus on the delta.

Coverage changes

If page coverage drops sharply, the scan is marked as degraded and resolved findings are not celebrated.