The instrument runs on us first
LFW Monitor scans lfw.com daily with the same engine, the same settings, and the same scoring it uses on client sites. The scores on our homepage are pulled from that scan — the live copy appears only when a completed scan exists, and there is no placeholder: if we haven't measured, the numbers simply aren't there.
Accessibility: full axe-core, WCAG 2.2 AA
Accessibility is audited with axe-core on every crawled page, against WCAG 2.2 AA. One honest detail worth stating: many tools also report Lighthouse's accessibility category alongside axe — but Lighthouse's accessibility audit is a subset of axe, so running both double-counts every issue and inflates the finding list. We run full axe, derive the accessibility score from it, and switch Lighthouse's accessibility category off.
Automated checks are the floor, not the ceiling: they catch the defects a machine can prove. Conformance work on client engagements adds manual testing on top.
Performance, SEO, best practices: Lighthouse
Performance, SEO, and best-practices scores come from Lighthouse, run page by page in a real headless Chrome. Scores are the standard 0–100 Lighthouse scale, so they're comparable to what any engineer would get running the same audit by hand.
Beyond the scores
The same crawl also checks the failures that don't show up in a score:
| Check | What it catches |
|---|---|
| Crawl & links | Broken links, redirect chains, unreachable pages |
| Content | Typos (with a per-site dictionary), stale copyright years |
| Integrity | Cloaking and injected spam or malware — the page is fetched as our own monitor, as desktop Chrome, as Googlebot, and as mobile Safari, and divergence between them is treated as the finding; the rendered DOM is also compared with the server HTML to catch script-assembled payloads |
| WordPress | REST-API media exposure, user enumeration, XML-RPC left open |
Findings are stateful — you see what changed
Findings carry a stable fingerprint across scans. A new scan updates the standing set rather than producing a fresh pile: findings are new, regressed, resolved, or silent because nothing changed. One rule matters more than it looks: a finding auto-resolves only when the page it lives on was actually reached this scan. A page that errors mid-crawl never quietly "fixes" its own issues.
What we publish, and what we never publish
For our own site we publish scores, page counts, and finding counts by severity. We never publish finding text — ours or anyone's. A public list of a site's unfixed defects is a map for attackers, which is also why client findings travel only through the client's own portal.