How LinkSweep scans
A LinkSweep scan discovers your pages from the sitemap and internal links, renders JavaScript in a real browser when a page needs it, runs 13 checks, then groups findings by severity and delivers them. Results are re-verified and reconciled against your sitemap to filter false positives, and every request is rate-limited per host so scanning stays polite. This page explains each step.
The scan lifecycle
- Discover. LinkSweep reads your XML sitemap and follows internal links to build the list of pages to check, up to your plan's page limit.
- Render when needed. Server-rendered HTML is read directly. If a page is a JavaScript shell, LinkSweep re-renders it in headless Chromium — see JavaScript & SPA crawling.
- Check. Each page is examined for broken links and images, SEO and accessibility issues, slow responses, security headers, mixed content, redirect chains and outdated content.
- Verify. Anything flagged is re-checked to rule out a transient failure, and on firewall-protected hosts the results are reconciled against your sitemap.
- Prioritise & deliver. Findings are grouped by severity and sent by email, webhook, CSV or PDF, depending on your plan.
How severity works
- Error
- Something is broken and visitors will hit it — a 404, a server error, an excessive redirect chain (5+ hops).
- Warning
- Should be fixed but not urgent — a 3–4 hop redirect, a missing meta description, a stale page, an accessibility gap.
- Info
- Context, not a defect — for example a link blocked by bot protection, reported so you can allowlist us rather than treated as broken.
How we avoid false positives
- Re-verification. A link that fails once is re-checked before it's reported, so a one-off timeout doesn't become a broken link.
- Sitemap reconciliation. On sites behind a firewall, flagged links are cross-checked against your sitemap so intermittent firewall responses don't turn valid pages into false errors.
- Date-pattern suppression. The outdated-date check ignores numbers that look like versions, prices or IDs, so it flags genuine stale dates rather than every number on the page.
- Bot-blocks are information. Links a firewall blocks are reported separately, with an allowlist path — never counted as broken.
Polite by default
LinkSweep respects robots.txt, meta-robots rules and crawl-delay. It caps how many requests it makes to any single host at once and adds an adaptive delay that grows if a host starts returning 429/503 responses (honouring Retry-After) and eases off once responses are clean again. Full technical detail is on the crawler specification.
See the methodology in action
Run a free whole-site scan and see how findings are prioritised. No credit card required.