Opens in a new tabSkip to content
Agent LighthouseAgent Lighthouse

    Searches the text of every published page. The evidence sources themselves are not in this index — search all of them on the trusted sources page.

    GitHub ↗
    Browse checks and page contents
    answer-readiness/unique-meta

    Unique meta per page

    What it checks

    AI crawlers use title and description pairs to distinguish between pages. Duplicate meta across pages causes agents to merge or skip content, meaning some of your pages will be invisible in AI-generated answers. Give each page a unique title and description.

    Why it matters

    A crawler or answer engine keys pages on their <title> + <meta name="description"> pair, so two pages sharing that pair are merged or one of them is dropped, making the dropped page unreachable in generated answers.

    Evidence

    • Google on duplicated descriptions: “Identical or similar descriptions on every page of a site aren’t helpful when individual pages appear in search results.” — developers.google.com/…/snippet (verified 2026-08-21)
    • Google on duplicated titles: “It’s important to have distinct text that describes the content of the page in the <title> element for each page on your site.” Boilerplate titles “that vary by only a single piece of information” are called out as bad — developers.google.com/…/title-link (verified 2026-08-21)
    • Both signals are real inputs to a named consumer: description is on Google’s supported-meta-tags list — developers.google.com/…/special-tags (verified 2026-08-21)

    Limits

    The documented Google effect is presentational (a less useful snippet, a rewritten title link), not exclusionary — nothing in the cited sources says a duplicate-meta page is merged, skipped, or dropped from the index. Google’s own duplicate handling is documented as canonicalization over page content and explicit canonical signals, not over meta pairs. No AI vendor documents any retrieval-time dedup keyed on meta tags. OpenAI’s crawler documentation mentions no HTML metadata — developers.openai.com/…/bots (verified 2026-08-21). Google’s AI-features page describes query fan-out and snippet controls with no reference to meta uniqueness — developers.google.com/…/ai-features (verified 2026-08-21). Legitimate duplicates also exist by design (paginated archives, faceted listings), so the current hard fail at priority high overstates a C-grade signal.

    How it scores

    Google documents that identical titles and descriptions across pages degrade how those pages are presented, which supports the convention. But no vendor documents title plus description as a deduplication or merge key. The “invisible in AI-generated answers” step of the claim has no source.

    Sources