GEO supplier accessibility acceptance

How should buyers set accessibility acceptance criteria for GEO content?

Use a six-part GEO accessibility acceptance record to define user journeys, standards, content, interaction, language and test evidence before approving content or technical changes.

Published by SALIENS · Published

Discoverable content can still exclude people

A page can be crawlable, factually accurate and well structured for discovery while remaining difficult for people to perceive, navigate or understand. A heading can look correct but lack semantic markup; an image can carry important information without an equivalent text alternative; a translated page can omit its language declaration; and a conversion path can fail for keyboard or assistive-technology users. GEO acceptance should therefore consider the human use of the page as well as whether systems can retrieve its information.

WCAG 2.2 organises accessibility under four principles: perceivable, operable, understandable and robust. Its success criteria include text alternatives, programmatically determinable information and relationships, keyboard operation, meaningful headings and labels, page language and compatibility with assistive technologies. These criteria provide a recognised reference point, but a general procurement checklist is not proof of conformance. The buyer must identify the applicable standard, level, law, policy and assurance route for its organisation and audience.

Create one six-part GEO accessibility acceptance record

Complete the record before content production or technical implementation. Apply it to a representative set of pages and journeys rather than an unspecified promise that the finished website will be accessible. Give every finding an affected URL or component, criterion, severity, evidence, owner, due date and retest result.

GEO accessibility acceptance record
Acceptance areaWhat to recordBuyer test
1. Users, scope and journeysTarget audiences, priority tasks, page and component sample, devices, browsers, assistive technologies, languages, exclusions and known constraints.Does the sample cover the content and journeys that matter, including a realistic bilingual path?
2. Standard and decision ruleApplicable WCAG version and level, organisational policy, legal or contractual requirements, permitted exceptions, severity model, acceptance authority and release threshold.Can reviewers tell what must pass, who may accept a residual issue and what evidence is required?
3. Content and structureHeadings, lists, tables, link purpose, reading order, page title, language attributes, plain-language review, meaningful images, alt text, captions and transcripts.Can the information be understood when presentation, colour or the image itself is unavailable?
4. Interaction and technical behaviourKeyboard path and focus, controls and labels, errors, status messages, zoom, reflow, contrast, motion, time limits, authentication and name-role-value information.Can priority tasks be completed without a mouse and without losing content, context or control?
5. Language and content equivalenceEnglish and Simplified Chinese page purpose, headings, instructions, labels, alternatives, error messages, links, metadata and declared language; approved differences and reviewers.Do both versions preserve the intended task and accessibility information rather than merely share a template?
6. Testing, evidence and closureAutomated, manual and assistive-technology tests; tester, date, environment, results, defects, fixes, retest, residual risk, approval and change trigger.Could an independent reviewer reproduce the result and trace every accepted exception or open issue?

Require more than an automated score

Automated tools are useful for repeatable checks such as the presence of some attributes, contrast calculations and selected markup problems, but they cannot decide whether alt text serves an equivalent purpose, whether a heading describes the section, whether the reading order makes sense or whether a real task is understandable. GOV.UK guidance combines automated and manual testing, and for government services also calls for assistive-technology and disabled-user testing. Its monitoring guidance distinguishes simplified sampling from detailed testing against the relevant success criteria.

Ask the supplier to state the method and limits of every test. Record the page version, browser, viewport, tool and rule set where relevant, and the assistive technology and task used for manual evidence. A pass from one scanner should not overwrite a manual failure, and a sample should not be presented as whole-site coverage. Where formal conformance or certification is required, commission the appropriate specialist audit and keep that decision separate from the GEO supplier's delivery evidence.

  • Use automation as one evidence source, not as the acceptance decision.
  • Test complete tasks and states, including errors, confirmation messages and dynamic changes.
  • Retain screenshots or recordings only where they add reproducible context; keep the underlying finding and criterion in text.
  • Retest the corrected page or component and record the exact version that closed the issue.

Integrate accessibility into the GEO workflow

Do not wait until publication. Add accessibility requirements to the brief, content template, design and component choices, translation instructions, implementation review and release gate. Content authors can provide descriptive headings, meaningful link text, concise instructions and useful alternatives. Developers can preserve semantic structure, language metadata, focus order, labels and status information. Reviewers can compare visible content, metadata, structured data and language alternatives without treating search markup as a substitute for accessible page content.

Keep the responsibilities explicit. A GEO supplier may recommend or implement content and technical improvements, while a design-system owner, development team, accessibility specialist, legal adviser or the buyer may own other decisions and evidence. If a dependency sits outside the supplier's control, record the owner and due date rather than marking the item complete. Accessibility acceptance should be linked to the same release, change-control and defect records used for the delivered page.

Run one representative bilingual acceptance exercise

Choose a journey that contains meaningful content, navigation, at least one image or data presentation, a link or form action, and both English and Simplified Chinese versions. Starting from the page entry, complete the task using keyboard navigation, review structure and alternatives, check zoom and reflow, confirm the declared page and passage language, inspect errors and status messages, and compare the two language versions. Link every finding to the release and repeat the affected checks after correction.

Treat the exercise as incomplete if the supplier cannot identify the tested version, if only a homepage was scanned, if the Chinese page was assumed to pass because it uses the same template, if important image text has no equivalent, if keyboard focus is lost, or if an unresolved issue has no named decision-maker. A documented limitation can be managed; an unbounded claim that everything is accessible cannot be verified.

Ask SALIENS for accessibility-aware GEO delivery evidence

SALIENS provides GEO audits, AI visibility monitoring, content and technical improvements, and ongoing GEO delivery. A scoped engagement can include buyer-defined accessibility acceptance criteria, page and journey sampling, semantic and content findings, English and Simplified Chinese equivalence review, source-linked recommendations, defect evidence and retest records alongside the wider GEO work.

Email info@oxfordintelligence.co.uk with your website, audiences, priority journeys, markets and languages, applicable accessibility standard or policy, current audit evidence, delivery owners, known issues and procurement timetable. SALIENS is operated by Oxford Intelligence Limited. A GEO review is not a substitute for a formal accessibility audit, certification or legal advice, and accessibility work does not guarantee crawling, indexation, AI mentions, citations, rankings, traffic or enquiries.

Sources

  1. W3C: Web Content Accessibility Guidelines (WCAG) 2.2 (checked 29 October 2026)
  2. W3C: Understanding WCAG 2.2 — explanatory guidance and testable success criteria (checked 29 October 2026)
  3. GOV.UK Service Manual: Testing for accessibility (checked 29 October 2026)
  4. GOV.UK Service Manual: Understanding WCAG 2.2 (checked 29 October 2026)
  5. GOV.UK: Accessibility monitoring — how we test (checked 29 October 2026)
  6. SALIENS: How should a GEO supplier service-level scorecard work?

Prepared with AI assistance using the linked sources and SALIENS service information.

Discuss your GEO project

info@oxfordintelligence.co.uk

SALIENS is a brand operated by Oxford Intelligence Limited.