Accessibility SEO audit
Accessible structure is machine-readable structure. The same semantic signals that help a screen reader understand your page — landmark roles, heading hierarchy, image alt text, labelled controls — also help crawlers, AI agents, and language models parse and extract your content. An accessibility audit is therefore a reliable pre-publish signal for crawl quality and AI visibility — not a parallel compliance track, but the same track.
Why accessibility and AI visibility share a root cause
Both assistive technology and web crawlers parse the HTML structure of a page, not the visual rendering. A heading that looks big because of CSS but is marked up as a <div class="big-text"> communicates nothing about document structure. A button that is a styled <div> is keyboard-inaccessible and semantically opaque. An image without alt text is invisible to any automated reader. In each case, the fix for accessibility is the fix for crawlability.
What an accessibility audit catches that a standard SEO tool misses
Heading hierarchy violations
Skipped heading levels (going from H1 directly to H3, or using multiple H1s) break the document outline that crawlers use to understand content relationships. Standard keyword tools do not flag this. axe-core and manual accessibility audits catch it immediately.
Missing or inadequate alt text
Images that communicate meaning — charts, diagrams, product photos, screenshots — are invisible to any automated reader without descriptive alt text. Google can sometimes infer image subject from surrounding context, but the information in the image is lost. For AI extraction, the rule is simpler: if the meaning is in the image and not in the text, it will not be cited.
Non-semantic interactive elements
<div> and <span> elements styled as buttons or links are invisible to keyboard navigation and to accessibility trees. They also confuse crawlers that try to map the interactive elements on a page. Real <button> and <a href> elements with visible focus states solve both problems simultaneously.
Landmark structure
HTML5 landmark elements — <header>, <main>, <nav>, <article>, <section>, <footer> — tell any automated reader which part of the page contains the primary content versus navigation, supplementary content, or site chrome. Without landmarks, a crawler must guess. With them, the primary content slot is explicitly labelled.
Colour contrast and legibility
WCAG 2.1 AA requires a contrast ratio of at least 4.5:1 for small text and 3:1 for large text. This does not affect crawl directly, but contrast violations are a reliable signal that the visual hierarchy of a page was built for looks over legibility — which correlates with other structural shortcuts that do affect crawl. An audit that finds contrast issues will usually find structure issues in the same pass.
Focus management and keyboard navigation
Keyboard-accessible navigation is required for WCAG 2.1 AA and Section 508. It is also the navigation model that many automated agents use when interacting with pages. Broken focus traps, missing focus indicators, and skip-navigation failures leave both human keyboard users and automated agents unable to reach parts of the page.
WCAG 2.1 AA and Section 508
Prompt Goblin audits to WCAG 2.1 Level AA — the standard required for US federal procurement (Section 508) and most enterprise B2B accessibility policies. The two standards are closely aligned: Section 508 incorporates WCAG 2.0 by reference and most organisations treat WCAG 2.1 AA as the practical target.
Government and institutional sites selling to the US public sector need Section 508 compliance as a contract prerequisite. The accessibility audit is the evidence artifact. We provide it as part of the scan report.
Accessibility audit checklist (abridged)
- One
<h1>per page; heading levels in strict order (no skips) - All landmark roles present:
banner,main,navigation,contentinfo - All informative images have non-empty, descriptive
altattributes - Decorative images have
alt=""(empty, not missing) - All interactive elements are reachable and operable by keyboard alone
- Visible focus ring on every focusable element (not just
:focus-visible) - All form inputs have associated
<label>elements oraria-label - Colour contrast ≥ 4.5:1 for body text; ≥ 3:1 for large text (18px+ bold or 24px+)
- No auto-playing audio without a pause control
- No flashing content faster than 3 Hz (seizure risk)
- Skip navigation link available for keyboard users
- Language declared on
<html lang="en"> - Page title (
<title>) is descriptive and unique
How Prompt Goblin runs the accessibility audit
The scan pipeline runs axe-core against the rendered page across multiple component states — not just a single viewport snapshot — and against real rendered HTML, not a source-code static analysis. Each finding is ranked by impact (critical / serious / moderate / minor) with a specific WCAG criterion, a code reference, and a fix prompt a developer can act on directly.
Accessibility findings are included in the same ranked task list as the citation and technical SEO findings. The engineer review step covers all three tracks in a single pass.
Frequently asked questions
Does fixing accessibility directly improve AI citations?
Fixing structural accessibility issues — heading order, landmark roles, semantic HTML — improves the machine-readability of the page. This raises the probability that a crawler correctly parses the content hierarchy, which is a prerequisite for citation. It is not a sufficient condition on its own; the three citation levers (mentions, rank, extractable content) still apply.
Do government sites need Product schema?
No. Service and government organisations correctly use Service, OfferCatalog, and GovernmentOrganization schema types. We never flag a service or government site for “missing Product schema” — that would be incorrect advice.
What tools does the audit use?
axe-core for automated rule checking (catches roughly 30–40% of WCAG issues mechanically), combined with manual review of heading structure, landmark roles, keyboard navigation, and colour contrast. Automated tools alone are not sufficient; manual review is required for a complete audit.