ASSESSMENT METHODOLOGY

How the A11yGo Accessibility Assessment Works

A transparent look at exactly what the automated assessment checks, how the website accessibility score is calculated, and — just as importantly — what it does not claim to measure.

The core principle: separate measurements

The assessment deliberately keeps four things separate: website accessibility (what was found), accessibility standards (WCAG / IS 17802 references), accessibility technology (any toolbar, widget or native controls detected on the page), and A11yGo enhancements (what A11yGo adds, measured independently). A website score is never changed by whether a toolbar is installed, and the score is never presented as a compliance score.

01 · WHAT IT MEASURES

Eight separate measurements

Each measurement is produced independently and reported separately.

A

Website Accessibility Score

Automated score (0–100) based only on detected characteristics.

B

Accessibility Issues

Normalized findings with severity, occurrences and impact.

C

Standards Mapping

WCAG 2.1 / 2.2 / WAI-ARIA references, only where verified.

D

Solution Detection

Vendor-neutral detection of any accessibility tool or controls.

E

Features Detected

Which accessibility features appear to be available.

F

A11yGo Coverage

A separate metric measuring A11yGo feature availability only.

G

Manual Review

Items automation cannot judge — clearly marked for human testing.

H

Recommended Actions

Prioritized remediation and enhancement guidance.

02 · PARAMETERS CHECKED

The 17 check categories

What the automated scanner examines on the page you submit.

Document & Page Structure

Missing or empty page title, document language, invalid language code, missing <main> / navigation landmarks, heading structure, skipped heading levels, duplicate IDs.

Images & Non-text Content

Images without alt text, empty alt, extremely long alt, SVG without an accessible name. Semantic correctness of alt text is flagged for manual verification.

Links

Empty links, image-only links, ambiguous text ("click here", "read more"), placeholder href="#" patterns.

Buttons & Interactive Controls

Buttons without accessible names, icon-only buttons, div/span used as interactive controls, missing roles/states.

Forms

Inputs, selects and textareas without labels, placeholder used as the only label — especially relevant for banks and NBFCs. Labels, required fields and error association.

Color & Contrast

Possible low-contrast inline styles, information conveyed only by color. Exact ratios are only claimed when they can be computed reliably.

Keyboard Accessibility

Positive tabindex, focus outline removal. Full keyboard operability is a manual review item.

Focus

Visible focus indicators, suspicious CSS that disables focus. Focus behavior inside dialogs is manual review.

WAI-ARIA Analysis

Invalid ARIA roles, broken aria-labelledby/describedby/controls references, redundant or inappropriate ARIA. WAI-ARIA is a technology, not a compliance framework.

Navigation

Skip links, navigation landmarks, consistent navigation, accessible names for navigation.

Text, Zoom & Reflow

user-scalable=no, restrictive maximum-scale, viewport restrictions that prevent zooming.

Motion & Animation

CSS animations, transitions, moving/flashing elements, prefers-reduced-motion support. Animation is flagged for review, not automatically as a failure.

Audio & Video

Video/audio elements, autoplay, captions <track>. Caption quality is manual verification.

Tables

Missing header cells, missing captions, complex tables — particularly relevant for financial institutions.

iFrames & Third-party Content

iFrames (payments, maps, chat, social), missing titles. Third-party content is reported, not automatically classed as inaccessible.

CAPTCHA

reCAPTCHA, hCaptcha and other implementations. Reported as "review recommended", not a failure.

PDFs & Documents

Links to PDF / Office documents. "Accessible document verification required" — link detection alone never claims PDF accessibility.

03 · SCORE METHODOLOGY

How the website accessibility score is calculated

The score is transparent and reproducible — never an arbitrary percentage.

Penalty model

The score starts at 100. Each normalized issue group subtracts a weighted penalty by severity:

SeverityPenalty / group
CRITICAL10 points
HIGH6 points
MEDIUM3 points
LOW1 point
INFO0 points (informational)

Normalization & frequency

Repeated identical issues are normalized into one issue group. For example, 150 images missing alt text is one "Missing alt text" group, not 150 penalties. A frequency factor (up to 2.5×) increases the impact when an issue is widespread:

OccurrencesFrequency factor
11.0×
101.5×
1002.0×
1,000+2.5× (capped)

Formula: penalty = severity weight × frequency factor, where frequency factor = min(1 + log₁₀(occurrences) × 0.5, 2.5). Final score = 100 − total penalty, clamped to 0–100.

Score bands

Bands describe the automated result — they are not WCAG conformance levels.

90–100
Excellent
75–89
Good
50–74
Needs Improvement
25–49
Poor
0–24
Critical

What the score is NOT

The score is an automated accessibility score, not a "WCAG compliance score", an "IS 17802 compliance score", or a legal determination of compliance. It measures accessibility characteristics detectable by automated checks only.

04 · STANDARDS MAPPING

Standards references policy

Every automated finding supports standards references where verified. The scanner maps:

  • WCAG 2.1 and WCAG 2.2 — success criteria are only attached when they are known and verified for the specific check (for example, missing alt text maps to WCAG 1.1.1; missing page title maps to WCAG 2.4.2).
  • WAI-ARIA — treated as a technology specification (name, role, value, relationships), not a compliance framework. Native semantic HTML is preferred.
  • IS 17802, EN 301 549, Section 508 — shown as "Standards mapping pending verification" because the scanner never invents clause numbers it has not verified.

No WCAG criterion number, IS 17802 clause or Section 508 reference is ever fabricated. If a mapping has not been verified, the report says so explicitly.

05 · VENDOR-NEUTRAL DETECTION

Accessibility solution detection & A11yGo coverage

Solution detection

The scanner detects whether the website already provides accessibility functionality — A11yGo, a third-party widget (UserWay, accessiBe, EqualWeb, AudioEye, One Click Accessibility, and others), custom/native controls, multiple solutions, or none at all.

Detection uses script URLs, CDN paths, DOM/CSS signatures and known vendor fingerprints. A website with no toolbar is never treated as an accessibility failure — it is simply reported as "no accessibility solution detected".

A11yGo Enhancement Coverage

If A11yGo is installed, a separate metric — A11yGo Enhancement Coverage — reports the percentage of applicable A11yGo features detected as available (e.g. text resize, high contrast, reduce motion, big cursor, dyslexic font, text-to-speech).

Coverage = detected-and-available features ÷ applicable features. It never changes the website accessibility score, and it is not presented as compliance. If A11yGo is not installed, coverage is simply "not available".

06 · MANUAL REVIEW

What automation cannot judge

The report clearly separates 18 manual review items from the automated score.

Keyboard-only operation
Screen-reader behavior
Meaningfulness of alt text
Quality of labels
Quality of error messages
Reading order
Cognitive accessibility
Caption accuracy
Audio description
Real-world user experience
Focus behavior
Dynamic content
Assistive technology compatibility
Exact color contrast values
Keyboard trap detection
Dialog & menu keyboard behavior
PDF / document accessibility
Motion & animation sensitivity

Assessment disclaimer

This assessment uses automated technical checks and identifies potential accessibility issues that can be detected automatically. Automated testing cannot identify every accessibility issue and does not replace manual accessibility evaluation, assistive technology testing or testing with people with disabilities. References to WCAG, IS 17802 and other standards are provided for informational and technical guidance. Results should not be interpreted as certification, legal advice or a determination of compliance.