A page-level claim and a site-level claim can reach different conclusions

What it is

The same checks can support a claim about “this page” or a claim about “this site.” If the home page passes and checkout fails, the page claim can stand and the site claim cannot. Scope of a conformance claim is the object the conclusion is bound to: a page, a set of pages, or a site (including every step of a process). Widening the scope can only expose more failures; it cannot hide a failed step inside an average.

Why it happens

Conformance is evaluated against a named set of web pages. Enlarge the set, and newly included pages can only add failures. Processes are stricter still: if any step of search → results → detail → pay fails, the process cannot be written into scope as conformant — the user cannot finish the task on “the page that passed.” The second layer is sampling. A site-level claim almost always rests on a representative sample (WCAG-EM's core, common, random, and complete processes). If templates outside the sample differ from those inside it, the claim does not cover the real site. The vaguer the scope, the more the claim reads as site-wide, and the easier it is to unscrew in an audit.

Studying it

Write the scope in WCAG-EM terms: basic attributes (host, authenticated or not), core processes, sampling method. Issue a conclusion for “sample pages only” and for “sample plus the complete checkout process,” and see whether the two flip.

Independent variables: scope (one page / sample set / site including processes). Dependent variables: conformance conclusion; the first step that falsifies the conclusion; failures reproduced on same-template pages that never entered the sample.

Do not let a crawler's “mean score per page” stand in for scope — an average has no scope.

Where it stops holding

Native apps, PDFs, and email templates are not “pages”; they need their own object, and a URL claim cannot carry them. Multiple hosts, an embedded admin, and inside versus outside a login wall are different sets. A page-level claim is legitimate; it must not be used in public material to imply the site. Narrowing scope must name the excluded URLs and the reason; quietly testing fewer pages is not a smaller scope, it is an incomplete claim.

Applying it

• The first sentence of every claim names the object: which URLs, whether authenticated views are in, which user processes must all be included.
• Site-level claims sample with WCAG-EM and bring complete core processes in. Marketing home pages alone are not allowed.
• Page-level claims stay in that page's delivery acceptance. When quoted, they must not be rewritten as site-level advertising.
• How to check: copy the scope out of the claim and retest one core process that sits outside it (usually checkout or publish). If that process fails while public language still says “the website conforms,” the object of the claim is wrong.

Related

• Same group: J1.10.2 Passing almost every check is not overall conformance · J1.10.3 Dynamically generated content must be retested after each update · J1.10.4 Third-party embeds usually sit outside the claimant's control
• Nearby: J1.12 Accessibility Statements and Conformance Documentation · J5.08 Limits of Automated Checking
• Search terms: scope of conformance claim · WCAG-EM · complete processes