Reviewing SEO Intelligence Findings

SEO findings are the concrete, prioritized issues StretchSearch detects on your indexed pages. They're produced by a deterministic engine — the same pages always yield the same findings — so results are reproducible and auditable, with no AI guesswork.

What the engine checks

StretchSearch analyzes each indexed page and the page set as a whole, producing typed, severity-ranked findings:

Per-page checks
- non_200_status (high) — the page returned an HTTP status other than 200.
- missing_title (high) — no <title> tag.
- title_length (low) — title outside the ideal ~10–65 characters.
- missing_meta_description (medium) — no meta description.
- meta_description_length (low) — meta description outside ~50–160 characters.
- missing_h1 (medium) — no <h1> heading.
- thin_content (medium) — under ~500 characters of extracted text.

Cross-page checks
- duplicate_title (medium) — a title shared by more than one page.
- duplicate_meta_description (low) — a meta description shared by more than one page.

The SEO health score

Alongside the findings, the engine computes an SEO health score from 0 to 1. It weights findings by severity (high counts most) against the number of pages, so one serious issue per page drives the score toward 0 and a clean site sits near 1. Use it as a single at-a-glance indicator of a project's SEO condition.

Read the overview

1. Open StretchSearch → SEO. The overview shows open findings, counts by severity (high/medium/low), the number of distinct finding types, and when findings were last generated.
2. Optionally filter the overview to a single project.

Work the findings list

1. Open Findings.
2. Filter by severity (high, medium, low) and status (default open).
3. Findings are ordered by severity, then most recent, so the highest-severity issues are always on top.
4. Open a finding to see its description, recommendation, and evidence (the URL and specifics like the offending status code or text length).
5. Fix the issue on the source page, then re-generate findings to confirm it clears.

How findings are generated and refreshed

Generation is a governed action performed by a platform operator. When findings are (re)generated for a project or crawl job, the prior open findings for that scope are marked superseded and a fresh set is inserted in one transaction. That means the open list always reflects the latest analysis, and you retain history via superseded findings.

A realistic example

After Northwind Coffee's pages are indexed, the SEO overview shows a health score of 0.62 with 4 high, 9 medium, and 6 low findings. Maya filters to high and finds two missing_title pages and one non_200_status page returning 404. She fixes the titles, removes the dead URL from the sitemap, and asks her operator to re-generate findings. The next overview shows the highs cleared and the score up to 0.81.

Tips

• Triage by severity: clear high findings (missing titles, non-200 pages) first — they carry the most weight in the health score.
• Treat duplicate_title/duplicate_meta_description as content-architecture signals; they often point to templating problems.
• thin_content uses extracted text length — if a page is content-rich but flagged, check that its main content is actually crawlable.
• Re-generate after a batch of fixes to supersede the old open findings and see your score move.

Troubleshooting

The findings list is empty. Either the pages are clean, or findings haven't been generated yet for this scope. Confirm pages are indexed, then have an operator generate findings.

A finding I fixed still shows open. Findings reflect the last generation. Re-generate to supersede the stale open finding.

Severity counts don't match what I expect. The overview counts only open findings for the scope. Superseded findings are retained but excluded from the open counts.