A slow page or unresponsive button can interrupt an enquiry. Measure important pages and test their mobile journeys before deciding where to invest. A technical score alone cannot tell you lost sales.
The Vodafone case on web.dev reports a 31% LCP improvement and 8% more sales in an A/B test. That is its result, rather than a forecast for your business.
Which three measures matter?
Core Web Vitals measure loading, responsiveness and visual stability. “Good” thresholds apply at the 75th percentile of visits, separately for mobile and desktop.
| Measure | Visitor experience | Good threshold |
|---|---|---|
| LCP | Main content appears | At most 2.5 seconds |
| INP | Page responds to interaction | At most 200 milliseconds |
| CLS | Elements stay in place | At most 0.1 |
An average can conceal poorly served visitors. Ask which pages and devices make up the result.
Real data or simulation?
PageSpeed Insights shows field data when a sample is available, followed by a simulated diagnostic. The former describes visitor experience; the latter helps find causes.
A standard Lighthouse page load cannot measure actual INP because it does not observe visitor interactions. Its blocking indicators can help diagnosis. Without field data, test real devices and consider collecting Web Vitals on your site.
Field reports aggregate a period rather than instantly reflecting a fix. Record change dates before comparing results.
Which pages should come first?
Choose pages receiving suitable traffic and contributing to an enquiry: service, campaign, product or contact. Inspect main content, navigation and forms.
Rank observed problems instead of ordering a redesign for a poor global score. An oversized hero image, third-party script and slow interaction require different corrections.
Test late banners and widgets. A button moving at the moment of a click can disrupt a visitor even when content loaded quickly.
What should the supplier deliver?
A diagnosis should identify responsible elements and estimate fixes. Request before-and-after measurements on the same pages in comparable conditions.
For images, adjust format and dimensions. Do not automatically delay the image providing the main content. For scripts, distinguish those needed by the journey from those removable or deferrable. Reserve space for elements arriving later.
Retest login, forms and measurement tools after changes. Breaking checkout or analytics while improving a score is not successful delivery.
How does this relate to commercial results?
Track enquiries and opportunities from corrected pages. Compare similar periods and sources. A changed campaign can alter conversion independently of speed.
For budgeting, use customer margin, enquiry-to-customer rate and several improvement assumptions. Label them as scenarios. Navigation or accessibility defects may merit fixing even without a precise financial estimate.
How do you keep the improvement?
Check pages after adding chat, a plugin or an image. Assign responsibility for regressions and corrections.
Technology affects maintenance but does not guarantee speed. The WordPress and Astro guide covers operating choices. The CRO guide inspects the wider journey, and website creation can include performance objectives in the brief.
