Three metrics, three different problems
LCP describes the appearance of the largest piece of content, INP describes responsiveness to interactions, and CLS describes unexpected layout shifts. The thresholds for a good result are up to 2.5 s, up to 200 ms and up to 0.1, respectively. The assessment uses the 75th percentile of observations; it is worth separating mobile devices and computers.
Source: web.dev: Web Vitals ↗
A high score on one metric does not mean that the other two are good. A quick view screen might have a button that responds with a delay, or a banner that moves the content just before you click it.
Look for the cause in a specific view
With LCP, check which element is being measured and when the browser learns about its resource. With INP, analyze interactions and long tasks performed on the main thread. With CLS, recreate the top-up moments of images, ads, fonts and messages.
“Reserve space for the product image before downloading and check that the purchase button does not change position.” This is more useful than the "improve page speed" command.
Separate diagnosis from user evaluation
A lab test helps reproduce the problem under controlled conditions. Real-world visit data shows the experiences of a broader group, but requires sufficient follow-up and time. The absence of field data does not mean a null result.
On this website, the Lighthouse score and available user data are shown separately. TBT from a laboratory test is not INP. You should not replace the name of a metric to make the report look more complete.
Also evaluate the implementation functionally
After optimization, check search, filters, forms and shopping cart. Removing a script can improve measurement while breaking an important action. Record what issue was resolved and what types of pages were audited.
A good performance score is a quality goal. It is not a contract for a specific position in Google or a substitute for the right offer and useful content.
FOR USE IN PRACTICE
Your checklist
- Check metrics separately for devices.
- Identify a specific element or interaction.
- Do not equate TBT with INP.
- After optimization, test important user actions.
Sources and news
Documentation reviewed on September 21, 2026. The examples in the article are illustrative - they do not describe audit results or client implementation.
Educational material developed with the help of AI based on the indicated sources. It does not replace an individual assessment of the website by a specialist. Editorial rules.