Site Audit: Errors, Warnings, and Notices Explained

How issues are categorized

StretchSEM's site audit sorts every finding into three severities by impact: errors (critical, fix first), warnings (important), and notices (polish). The Site Health score weights them accordingly — errors most, notices least. Use this reference to understand and resolve each issue type.

Errors (critical)

Broken links (4xx/5xx) Internal or external links that point to a URL returning a client or server error. Why it matters: broken links waste crawl budget, frustrate users, and leak link equity. Fix: update or remove the link, or restore the target page. The report samples the broken URL and the status it returned (e.g. → 404).

Pages returning non-200 Indexed pages that don't return HTTP 200 OK (for example a 301, 404, or 500). Why it matters: search engines may drop non-200 pages from the index, and users may hit errors. Fix: ensure important pages return 200; redirect or remove ones that shouldn't be indexed.

Missing <title> Pages with no title tag. Why it matters: the title is the single strongest on-page ranking signal and the clickable headline in search results. Fix: add a unique, descriptive title (ideally ≤65 characters) to every page.

Warnings

Orphan pages Indexed pages with no internal links pointing to them. Why it matters: orphaned pages are hard for crawlers and users to discover, so they rarely rank. Fix: link to them from relevant, well-linked pages (hubs, navigation, related content).

Missing meta description Pages with no meta description. Why it matters: while not a direct ranking factor, the meta description is the snippet under your title in search results and strongly affects click-through. Fix: write a compelling ~120–155 character description featuring the target keyword.

Missing <h1> Pages with no H1 heading. Why it matters: the H1 signals the page's primary topic to search engines and assistive technology. Fix: add one clear H1 per page that includes the primary keyword.

Thin content Pages with very little extracted text (under ~300 characters). Why it matters: thin pages rarely satisfy search intent and can drag down site quality signals. Fix: expand with genuinely useful content, or consolidate/remove low-value pages.

Non-HTTPS / mixed content Pages served over plain HTTP. Why it matters: HTTPS is a baseline trust and ranking signal; mixed content triggers browser warnings. Fix: serve all pages over HTTPS and update internal references.

Notices

Duplicate titles Multiple pages sharing the same title. Why it matters: duplicate titles confuse search engines about which page to rank and can cause the wrong page to surface. Fix: give every page a unique title; this often overlaps with keyword cannibalization (see the glossary).

Pages reached via redirect Link targets that resolve through a redirect. Why it matters: redirect hops add latency and dilute link signals slightly. Fix: link directly to the final URL.

Title too long Titles over ~65 characters. Why it matters: long titles truncate in search results, hiding your message. Fix: tighten titles so the key phrase stays visible.

A suggested fix order

1. Broken links and non-200 pages — user-facing and heavily weighted.
2. Missing titles — highest-impact on-page signal.
3. Orphan pages — reconnect them to your internal link structure.
4. Missing meta/H1, thin content, mixed content — quality and crawlability.
5. Duplicate titles, redirects, long titles — polish that lifts the score's last few points.

Tips

• The report is pre-sorted by severity and count — work top-down.
• Each finding includes a sample of URLs; start there.
• Re-run the audit after each batch to watch Site Health climb.
• Duplicate titles are often a symptom of cannibalization — check the Domain report's cannibalization section too.

FAQ

Does fixing notices meaningfully change my score? A little — notices are lightly weighted. Prioritize errors and warnings for the biggest gains.

Why is the same issue flagged as both a duplicate title and cannibalization? Duplicate titles are a metadata problem; cannibalization is a keyword-targeting problem. They frequently co-occur when two pages chase the same term.