Key answers are visible in HTML and self-contained

Microsoft's guidance for AI search answers is blunt about where content gets lost: "Don't hide important answers in tabs or expandable menus: AI systems may not render hidden content," keep core information out of PDFs, and avoid key facts that exist only inside images. Bing also recommends self-contained phrasing — "sentences that make sense even when pulled out of context" — because assistants assemble answers from extracted passages. A pricing table that exists only as a screenshot, or a core FAQ answer locked in a PDF, is invisible to the systems deciding whether to cite you. Note what this check is NOT: Google explicitly debunks "chunking" content for AI ("There's no requirement to break your content into tiny pieces … There's no ideal page length"), so this check imposes no word counts, paragraph lengths, or section sizes.

Put core answers in served HTML text. If pricing, specs, or FAQ answers live only in a PDF, an image, or a client-rendered tab panel, add an HTML rendering of the same facts on the page. Prefer self-contained sentences for key claims ("AuditBuffet's Stack Scan runs 25 checks in about 30 seconds" survives extraction; "It runs them in about 30 seconds" does not). Server-rendered disclosure widgets like <details> are fine — the text is in the HTML.

// Instead of an image-only pricing table:
<table>
  <caption>Pricing</caption>
  <tr><td>Free</td><td>Stack Scan, 25 checks</td></tr>
  <tr><td>Pro, $9/mo</td><td>All 117 audits</td></tr>
</table>

• ID: answers-visible-and-self-contained

• Severity: low

• What to look for: On marketing and content pages, identify the key informational content a visitor (or an AI assistant) would come for: pricing, core feature claims, FAQ answers, specs, how-it-works. For each, determine the format it ships in. Count the key information items found and classify each as: (a) text in served HTML, (b) only inside an image (screenshot of a table, infographic with no text equivalent), (c) only inside a linked PDF, (d) only inside client-rendered UI whose text is absent from the served HTML (a tab panel populated by JS state, not a server-rendered <details>). Before evaluating, quote one example of the page's most important claim and note its format.

• Pass criteria: Count all key information items identified. 0 items are image-only, PDF-only, or client-rendered-hidden. Server-rendered collapsed UI (<details>/<summary>, CSS-hidden text that exists in the HTML source) PASSES — the text is in the served document. Report even on pass: "X key information items checked — all present as HTML text. (No word-count or section-length criteria apply: Google explicitly debunks chunking content for AI.)"

• Fail criteria: At least 1 key information item exists ONLY as an image, ONLY in a PDF, or ONLY in client-rendered UI whose text never reaches the served HTML. Report: "X of Y key information items are not extractable as HTML text: [item — format]".

• Do NOT pass when: the pricing table or core FAQ answers are screenshots, even if alt text summarizes them — a one-line alt is not the answer content; or when the only how-it-works explanation is a downloadable PDF.

• Do NOT fail for: long paragraphs, long sections, long pages, or absent subheadings. Google's guidance is explicit that there is no ideal page length and no requirement to fragment content for AI. Content entirely absent from served HTML because the whole page is client-rendered belongs to content-server-rendered — cross-reference it rather than failing the same root cause twice.

• Skip (N/A) when: No marketing or content pages (API-only or utility project).

• Detail on fail: "Pricing exists only as a screenshot (pricing.png) — per Microsoft's AI-search guidance, key info locked in images is not available to AI systems assembling answers; add an HTML table." or "2 of 6 key items (FAQ answers, integration steps) live only in client-rendered tab panels with no text in served HTML — Bing: 'AI systems may not render hidden content.'"

• Remediation: Render the same facts as HTML text alongside (or instead of) the image/PDF/JS-only widget, and phrase key claims so they stand alone when extracted:

  // Self-contained: survives being quoted out of context
  <p>AuditBuffet's free Stack Scan runs 25 security, legal, and abuse checks
  in about 30 seconds, without your code leaving your machine.</p>