Method
What we measure, and why it matters
A good website is found, opens quickly, works for everyone, can be trusted and leads visitors to get in touch or book. These reports follow the parts of that we can measure reliably and repeat over time.
1. Speed: does the page open quickly?
Speed is the measure with the clearest link to results. Visitors give up on slow pages, most of all on a phone with patchy reception, and Google uses speed as one of its ranking signals through its Core Web Vitals: how quickly the main content appears (LCP), how quickly the page reacts to a tap (INP) and how still it stays while loading (CLS). Speed alone will not put a page at the top of Google, since relevant content matters far more, but a slow page loses visitors who already found it.
The speed score is Google Lighthouse's Performance score. It weighs five lab measurements: Total Blocking Time 30%, Largest Contentful Paint 25%, Cumulative Layout Shift 25%, First Contentful Paint 10% and Speed Index 10%. Total Blocking Time is the lab stand-in for INP, which can only be measured with real visitors.
| Measure | Good | Poor |
|---|---|---|
| Main content appearsLargest Contentful Paint (LCP) · Core Web Vital | up to 2.5 s | over 4.0 s |
| Page ignores tapsTotal Blocking Time (TBT) · lab stand-in for INP | up to 200 ms | over 600 ms |
| Page jumps while loadingCumulative Layout Shift (CLS) · Core Web Vital | up to 0.10 | over 0.25 |
| First thing appearsFirst Contentful Paint (FCP) | up to 1.8 s | over 3.0 s |
| Page looks completeSpeed Index | up to 3.4 s | over 5.8 s |
| Server answersTime to First Byte (TTFB) | up to 800 ms | over 1.8 s |
2. Findability: can people find it?
Lighthouse's SEO score covers the technical basics: a page Google is allowed to read, a title and description, working links, image descriptions, the right language versions and a canonical address. A score of 100 means nothing is in the way, not that the website ranks well. On top of that the reports count the pages listed in the sitemap, the languages offered and the structured data that tells Google what the business is (a hotel, a restaurant, its address and opening hours).
For a local business like a hotel or restaurant, the biggest factor sits outside the website: the Google Business Profile, with its primary category, photos, opening hours and, above all, reviews. Surveys of local search specialists consistently rank the profile and reviews first, with the website's own content and links close behind. The truest measure of findability is therefore Google Search Console: how often the website appears in search results, how many people click and for which searches. That is the next data source to add.
More and more people ask an AI assistant instead of searching. Lighthouse 13 added an experimental Agentic Browsing score that checks whether such assistants can read and use a page. It is still under development, so the reports show it but do not weigh it as heavily as the four main scores.
3. Accessibility: can everyone use it?
About one in six people worldwide lives with a significant disability, and many more use a website with tired eyes, in bright sunlight or with one hand. The Accessibility score checks contrast, image descriptions, labels on buttons and form fields, page structure and tap target sizes. Since 28 June 2025 the European Accessibility Act requires websites that sell to consumers in the EU, including online booking, to meet WCAG 2.1 level AA (micro-enterprises are exempt).
Automated checks only catch part of the problems. A score of 100 is a good sign, not proof of an accessible website; keyboard use and screen readers still need to be tried by hand.
4. Trust and safety: is it built well?
The Best Practices score checks for a secure connection, errors in the background, outdated code and images shown at the wrong size. The reports add what a fresh phone browser sees without clicking anything: how many of five recommended browser protections are switched on, how many outside companies are contacted, how many cookies are set before the visitor agrees, and how many errors appear while loading. Fewer outside companies means a faster page and a simpler privacy policy.
5. Footprint
Every megabyte sent costs the visitor time and data and uses electricity. The reports show the home page's total download size and estimate its CO₂ per 1,000 visits with version 3 of the Sustainable Web Design model (0.81 kWh per gigabyte at 442 g CO₂ per kWh). It is an estimate for spotting trends, not an exact footprint.
The overall score
The overall score puts everything above into one number from 0 to 100, so different versions of a website can be compared at a glance. Each part counts with a fixed weight, and the phone and computer results count equally:
| Part | Weight |
|---|---|
| Speed | 30% |
| Findability | 20% |
| Accessibility | 20% |
| Trust & safety | 15% |
| Privacy and safety checks | 15% |
Speed weighs most because it affects every visitor on every page. The privacy and safety checks score the share of the five recommended protections that are switched on, and lose 20 points for every outside company and every cookie set before the visitor agrees and 25 points for every error while loading. When a measurement did not record those checks, it uses those of the nearest measurement of the same version, and if there is none, the other parts share their weight. The experimental AI-assistant score is left out until Lighthouse finishes it.
Versions of a website from before these reports started were never measured. They are included with every value set to 0, so they appear in the overview, and are shown as gray squares in a separate strip on the left of the score charts. They are left out of the speed charts and every before-and-after comparison, where 0 would wrongly read as instant.
How the measurements are taken
- Each measurement runs Google Lighthouse, the engine behind PageSpeed Insights, on the home page three times on an emulated phone and three times on a computer, and keeps the middle result. Every other page is measured once on a phone.
- The phone test imitates a mid-range Android phone on a slow 4G connection, so real visitors on good connections usually see shorter times. What matters is the trend: the same test, run the same way, over time.
- Scores move a few points between runs even when nothing changed. A change of 5 points or less in a single measurement is noise; a change that holds over several measurements is real.
- Changes to a website are logged as milestones, so each one can be compared with the measurement just before and just after it. Measurements of a website on a separate test address are shown as a dashed line.
Not measured yet
- Search results from Google Search Console: impressions, clicks, average position.
- Real-visitor speed from the Chrome UX Report, once a website has enough visitors for Google to publish it.
- Google Business Profile: rating, number of reviews, views and direction requests.
- Results: contact and booking requests coming in through the website.
Sources
- Web Vitals: thresholds and why they matter (web.dev)
- Lighthouse performance scoring and metric weights (Chrome for Developers)
- Understanding page experience in Google Search results (Google Search Central)
- European Accessibility Act, Directive (EU) 2019/882 (EUR-Lex)
- Local Search Ranking Factors survey (Whitespark)
- Estimating digital emissions, Sustainable Web Design model