Working out the focus site’s rating…
The focus site’s figures, with its position among the websites that have each measure. A green top edge marks a top-three place.
Pick a measure. Sites doing better than the focus site are listed above it. Tap any site to compare it with the focus site on every measure.
Click a column heading to sort. Green, amber and red follow Google's own pass marks for each measure. The focus site is highlighted and stays in view. Gold, silver and bronze mark the top three in each column among the sites shown (ties share a medal).
The overall rating is out of 100. It weights six themes by how much they matter, not by how many columns each has. Each measure is put on a fixed 0–100 scale, so a score means the same whichever sites are shown. Measures are averaged within their theme, then themes are combined by weight. A site is rated only when it has both the phone speed test and real-visitor data. If another check could not be run on a rated site (for example, its bot protection blocked it), the gap is filled with the middle value of the other rated sites, so a missing check neither adds nor removes points; the site's Details say how many gaps were filled. A check counts only once at least 80% of rated sites have a result for it. Visitor numbers, site size and phone share describe the business, not the website, so they are not part of the rating. Open a site's Details to see its score for each theme.
Loading…
Every column explained in plain terms: what it measures, why it matters, who leads, and what the leading websites do differently from the focus site. Open a measure to see its chart. The figure on the right is the focus site’s, with its position among all websites that have the measure.
No single test tells the whole story, so the benchmark combines lab tests (repeatable, good for finding causes), real-visitor data (what people actually got), checks of quality and trust, and context about each business. All were run or collected on 28 Oct 2026 unless noted.
| Tool | What it checks | Why it is included | Limits |
|---|---|---|---|
| Google Lighthouse 13 | Speed, accessibility, best practice and search basics of each homepage (and UK attraction tickets pages) on a simulated phone and a desktop. | The industry-standard lab test; the details inside it explain why a page is slow. | One page per site; scores swing a few points between runs (the middle of three is kept); harsher than real life. |
| Chrome UX Report | What real Chrome users on phones experienced across each whole site in September 2026, plus UK popularity bands. | It is real-world, and it is what Google uses in search ranking. | Chrome users only; needs enough traffic (two small sites have none); monthly, so it lags changes by weeks. |
| axe-core 4.10 | Homepage accessibility against WCAG 2.2 A and AA. | The most widely used automated accessibility engine; it names each fault. | Automated checks find about a third to a half of problems; some sites blocked it. |
| Browser check for this benchmark (Playwright with Chrome) | Structured data, languages, platform and booking links on each homepage. | No public service checks structured data in this way. | One visit from one UK location; sites may behave differently elsewhere or later. |
| Mozilla HTTP Observatory | Security headers that protect visitors (A+ to F). | A free, respected and repeatable security grade. | Checks settings, not whether a site has actually been attacked. |
| Green Web Foundation and CO2.js | Green hosting, and estimated carbon per homepage view. | The standard open tools for website carbon. | Estimates from page size. Sites behind a delivery network such as Cloudflare show as not known, because the check sees the network rather than the site’s own server. |
| ALVA 2025 | Visits to each attraction in 2025. | Context: the size of the business behind each website. | Members only; many sites are not in it. |
| Industry benchmarks | State of Museum Digital: 2026 Benchmark (Urban Insight and Pantheon) and UK averages from the Chrome UX Report, shown as italic average rows. | Reference points beyond the sites in this table. | The museum report names its tools (Google PageSpeed Insights, axe-core) but not its test dates, settings or museum list, and covers US museums only, so compare loosely. |
The figures live in a separate data file (data.json), one record per website, and the table is built from a list of columns. A new measure is one new field per site plus one column entry, so the table can grow without being redesigned. Use "Show columns" to hide groups you don't need.
Still to add:
C:\dev\website-monitor\audits\lighthouse\benchmark-20261028\ (run.sh reruns the test; merge.py rebuilds the data file).