How the pages are chosen
We start with your homepage, then collect same-site links from it and the URLs in your sitemap. To avoid testing ten blog posts that share one layout, we group addresses by their path pattern — /services/employment-law and /services/tax-advice count as one template — and pick one page per template, preferring pages linked from the homepage. Pages that robots.txt disallows, files such as PDFs and other hosts are skipped. Up to six pages are tested.
Each page then gets its own Google Lighthouse performance run on a simulated mid-range phone with throttled 4G, one after another, so they do not slow each other down. The whole scan has a time budget of four minutes; if it runs out, the pages measured so far are reported.
What the report shows
- A table of every tested page sorted slowest first, with its performance score as a bar.
- Largest Contentful Paint, Total Blocking Time, Cumulative Layout Shift, First Contentful Paint and Speed Index for each page.
- Page weight and number of requests, which usually explain the differences.
- The biggest Lighthouse opportunity per page, with a concrete fix for slow pages.
- The median score across pages and the spread between the fastest and the slowest.
Why test more than the homepage
Search engines assess pages individually, and many visitors arrive directly on a service page, an article or the contact page from a search or an advertisement. Those templates often carry extra weight the homepage does not: an embedded map, a video, a slider, a booking widget or a page builder that loads every module everywhere. A fast homepage can hide a slow contact page — the one page where a delay costs you an enquiry.
Lighthouse scores vary by a few points from run to run. Look at large gaps and at the metrics behind the score rather than single-point differences.
Fixing the slowest page first
Start at the top of the table. Compare the slowest page with the fastest one: the difference in page weight and requests usually points straight at the cause, such as an uncompressed hero image, a map or video embed that could load on click, or a plugin that adds scripts to every page. Fix one template and every page built on it improves at once, then run the test again to confirm the gain.