How we work · Monitoring tools
LFW Monitor
Daily checks on the things that fail quietly. Our own results, published.
Recorded findings for this website
Open findings currently recorded for this site, across 62 completed scans.
- Content
- 81Open findings
- Website security
- 68Open findings
- SEO
- 33Open findings
- Links
- 21Open findings
- Accessibility
- 15Open findings
- Performance
- 5Open findings
- Email & TLS
- 4Open findings
- Best practices
- 1Open findings
Open findings are shown as counts, deliberately: publishing an unfixed issue's details would hand out a map. LFW reviews findings and shares relevant details and next steps in the portal.
Read resolved examples
Real examples, caught here and fixed here
- serious Elements must meet minimum color contrast ratio thresholds (fixed on 27 pages)
- serious Phone visitors download desktop-sized images (fixed on 13 pages)
- serious Dead link: https://lfw.com/cdn-cgi/l/email-protection/
209 findings caught and resolved on this site so far. The engine finds real things; the operator closes them.
Latest completed scan of this website, September 13, 2026. Rounded automated averages. How to read these numbers
Discover the public pages
Sitemaps and internal links build a bounded URL inventory. Review indexing, redirects, metadata, and possible gaps in navigation.
Check the visitor experience
A selected page sample gets desktop and optional mobile checks. Configured journeys verify visible text and links without submitting anything.
Know what was covered
Separate inventory and browser counts show what completed, failed, or was omitted. LFW reviews findings and shares useful next steps in your portal.
Reading the numbers
Reading the numbers
Four scores, four different kinds of measurement. Here is how each one is produced, what good looks like, how ours got where they are, and what quietly moves them.
How they are measured
Performance, SEO, and Best practices are Google Lighthouse category scores. Browser audits use a selected page sample, with separate desktop and optional throttled mobile performance measurements. Each score is a rounded average of successful measurements; checks that fail to complete are excluded, so coverage can differ by metric.
Accessibility is our own score, based on axe-core checks for WCAG 2.2 AA and accessibility best practices. Failing rules cost points by impact (critical 10, serious 6, moderate 3, minor 1), with an additional penalty for repeated elements. The page scores are averaged and rounded.
How to use the results
Use these scores to spot changes and decide what to investigate. Lighthouse performance is a lab measurement under a simulated device and connection; it does not measure how many visitors leave. Compare results alongside the scan date, coverage, and individual findings.
An accessibility average of 100 can still include failing rules because of rounding. It is not a certification that every page is accessible. Review the findings, then check keyboard operation, screen-reader use, and other interactions that automated checks cannot fully assess.
Our own record
In the history shown here, desktop performance starts at 83 on August 2, reaches a low of 81 on August 3, and ends at 100 in the latest recorded scan. Changes in page coverage and test conditions can affect the average.
Show the record as a table
| Day | Performance | Accessibility | SEO | Best practices |
|---|---|---|---|---|
| 2026-08-02 | 83 | 100 | 97 | 100 |
| 2026-08-03 | 81 | 100 | 97 | 100 |
| 2026-08-04 | 81 | 100 | 97 | 100 |
| 2026-08-05 | 81 | 100 | 97 | 100 |
| 2026-08-06 | 81 | 100 | 97 | 100 |
| 2026-08-07 | 81 | 100 | 97 | 100 |
| 2026-08-08 | 81 | 92 | 97 | 100 |
| 2026-08-09 | 82 | 93 | 97 | 100 |
| 2026-08-10 | 100 | 100 | 100 | 100 |
| 2026-08-11 | 100 | 100 | 100 | 100 |
| 2026-08-12 | 100 | 100 | 100 | 100 |
| 2026-08-13 | 100 | 100 | 100 | 100 |
| 2026-08-14 | 100 | 100 | 100 | 100 |
| 2026-08-15 | 100 | 100 | 100 | 100 |
| 2026-08-16 | 100 | 100 | 100 | 100 |
| 2026-08-17 | 100 | 100 | 100 | 100 |
| 2026-08-18 | 100 | 100 | 100 | 100 |
| 2026-08-19 | 100 | 100 | 100 | 100 |
| 2026-08-20 | 100 | 100 | 100 | 100 |
| 2026-08-21 | 100 | 100 | 100 | 100 |
| 2026-08-22 | 100 | 100 | 100 | 100 |
| 2026-08-23 | 100 | 100 | 100 | 100 |
| 2026-08-24 | 100 | 100 | 100 | 100 |
| 2026-08-25 | 100 | 100 | 100 | 100 |
| 2026-08-26 | 100 | 100 | 100 | 100 |
| 2026-08-27 | 100 | 100 | 100 | 100 |
| 2026-08-28 | 100 | 100 | 100 | 100 |
| 2026-08-29 | 100 | 100 | 100 | 100 |
| 2026-08-30 | 100 | 100 | 100 | 100 |
| 2026-08-31 | 100 | 100 | 100 | 100 |
| 2026-09-01 | 100 | 100 | 100 | 100 |
| 2026-09-02 | 100 | 100 | 100 | 100 |
| 2026-09-03 | 100 | 100 | 100 | 100 |
| 2026-09-04 | 100 | 100 | 100 | 100 |
| 2026-09-05 | 100 | 100 | 100 | 100 |
| 2026-09-06 | 100 | 100 | 100 | 100 |
| 2026-09-07 | 100 | 100 | 100 | 100 |
| 2026-09-08 | 100 | 100 | 100 | 100 |
| 2026-09-09 | 100 | 100 | 100 | 100 |
| 2026-09-10 | 100 | 100 | 100 | 100 |
| 2026-09-11 | 100 | 100 | 100 | 100 |
| 2026-09-12 | 100 | 100 | 100 | 100 |
| 2026-09-13 | – | – | – | – |
What to watch for
- A new hero image uploaded at full size. Phones download it whole.
- A video or map embed dropped in without a reserved box. The page jumps when it loads.
- A plugin update that adds a script to every page. Performance drops overnight and nobody changed anything they can name.
- An email policy still at p=none. Anyone can send mail as you and have it delivered.
- Files parked in the media library. They are public the moment they are uploaded.
- A certificate that renews next month. Almost always automatic. Worth confirming once.
What is checked, how scores work & scope
The things that fail quietly
A certificate expiring in three weeks. A plugin update that hid a form label or pushed a page sideways. A page that quietly stopped being indexable. None of these announce themselves. They just accumulate until a visitor, or a board member, finds them first.
LFW Monitor runs scheduled checks and shows what is new, what returned, and what was resolved. Coverage is recorded separately for the URL inventory and the heavier browser measurements. A failed or skipped check is not a clean result.
What's checked
- A public URL inventory from sitemaps and internal links, with explicit page limits and exclusions
- Sitemap conflicts: noindex pages, redirects, errors, duplicate URLs, and mismatched canonical addresses
- Missing or duplicate titles and descriptions, plus editorial review signals for brief metadata and possibly unlinked pages
- Accessibility against WCAG 2.2 AA using axe-core on the selected browser sample
- Desktop and optional mobile performance, layout stability, and repeated response-time measurements when a slow response is observed
- Read-only browser checks for sideways overflow, failed resources, script errors, visible form labels, and configured page-to-page journeys
- Broken links, internal and external
- Email deliverability (SPF/DKIM/DMARC) and the domain's TLS certificate
- Integrity: cloaking and injected spam, caught by comparing what different visitors are served
- WordPress exposure, when the site runs WordPress: REST API leaks, outdated plugins, common misconfigurations
Intentional noindex pages and temporary redirects can be recorded as exceptions. Short copy is a prompt for review, not an automatic SEO failure. Possible orphan pages remain provisional when the public link graph is incomplete or links depend on JavaScript.
The public URL inventory starts with a 400-page limit, adjustable to the agreed scope. Customer browser checks follow the agreed page scope. On LFW.com, all discovered public HTML pages receive desktop and mobile performance checks, including blog articles. Image-size checks also flag heavy original and responsive images.
LFW can configure up to ten read-only journey checks, such as confirming that your quote page shows its next step and links to booking. A failed or skipped check retains its earlier findings until that check completes or the finding is reviewed.
Journey checks look for expected visible text and links. They do not submit forms, send messages, complete payments, or prove delivery. Response timings come from our monitoring location, not from a verified Google, OpenAI, or other crawler connection.
Everything above the last line is platform-independent: it tests what your site actually serves a visitor, so it works the same on Drupal, on WordPress, or on a static build. That matters for a public agency, where the CMS was chosen years ago and isn’t the thing up for discussion.
The full method is documented in the open on the methodology page: which tools, which settings, what auto-resolves and what never does.
Try the first pass, free
The full monitoring service runs daily on infrastructure we operate. You can run a real, public-information first pass on any domain right now, free, no account needed.
Run the free audit →The free audit is a public-information first pass. Managed monitoring adds a scheduled URL inventory, a selected browser sample, configured read-only journeys, and a record of changes. Authenticated pages and form submissions are outside these automated checks; any additional testing is agreed separately.
Pricing
Quoted to the scope of your site, as its own line item or within a Web Operations Retainer. Either way the number is in writing before anything starts. The free audit above has no gate at all.
See it pointed at your own site.
Twenty minutes, no sales pressure. Or start with the free audit and get a real first pass on your site today.