Key Takeaways
- Core Web Vitals cover loading, responsiveness and visual stability.
- Check field data where available; a lab score is a diagnostic, not a ranking guarantee.
- Test the customer journey after optimising scripts, images or caching.
A speed report should tell you what a visitor struggles to do, not just whether a score is green. Start with important landing pages and the enquiry or purchase journey. A fast homepage does not establish that the rest of the site works well.
Understand the three measures
Google's Web Vitals documentation sets these good-experience thresholds, assessed at the 75th percentile of visits, with mobile and desktop considered separately:
| Metric | What it measures | Good threshold |
|---|---|---|
| Largest Contentful Paint (LCP) | Loading of the main visible content | 2.5 seconds or less |
| Interaction to Next Paint (INP) | Responsiveness to interactions | 200 milliseconds or less |
| Cumulative Layout Shift (CLS) | Unexpected visual movement | 0.1 or less |
These are not predictions of sales or positions. A small site may lack enough field observations for a URL-level report. Missing field data is not a pass or a fail, and a single lab run cannot substitute for real-user results.
Make a repeatable test record
Choose the homepage, a priority service page, an article and a form or product page. Note the date, device, test tool and whether the result is URL-level or origin-level data. Repeat an unusually poor run before treating it as a persistent fault.
Use lab diagnostics to investigate a particular resource or interaction. Check the page on a phone as well: does the main image appear late, does a button move as a banner loads, or does opening a menu stall? Describe the problem in terms the owner can verify.
Fix the bottleneck you actually observed
For loading delays, separate server response time from browser rendering. Check whether an oversized image, font or blocked resource delays the main content. Do not lazy-load the main visible image simply because the setting is available; images further down the page can use a different strategy.
For interaction problems, inspect the work performed when the user clicks or types. A third-party widget, long-running script or expensive page update may be responsible. Removing essential form validation to improve a metric is not an acceptable repair.
For layout shifts, reserve space for images and embeds and examine late-loading fonts or banners. Check the actual movement around controls. A change that prevents a jump on desktop may still need a different layout on a small screen.
Retest what the business depends on
Caching and script changes can affect consent, analytics, menus and checkout. Use staging for risky changes, preserve a rollback route and test both success and failure states. Private and personalised pages need appropriate cache behaviour, not indiscriminate caching.
The WordPress performance guide covers platform-specific checks. The maintenance guide helps assign responsibility for keeping the improvement working after updates.
Connect performance work to the right decision
Google's page-experience guidance makes clear that good experience does not guarantee top rankings. Treat speed as one part of a useful, accessible site, alongside relevant content and working enquiry routes.
Our WordPress speed audit starts with diagnosis. Share the slow pages and affected tasks, including what happens on mobile. If visits arrive but enquiries do not, use the conversion diagnosis guide rather than assuming every commercial problem is caused by milliseconds.
Frequently Asked Questions
Does a Lighthouse score of 100 guarantee better rankings?
No. A lab score is one test under specified conditions. Search relevance, content and other factors remain important, and field experience can differ.
What if PageSpeed Insights has no field data?
Record the gap rather than treating it as healthy or broken. Use repeatable diagnostics and real-device checks while assessing whether suitable real-user measurement is available.
Should we delay every script to improve speed?
No. Identify unnecessary work and test changes carefully. Forms, consent, navigation and payment flows must remain functional and correctly measured.
