Someone opens a project on their phone, tries to read the details and taps the enquiry link. A speed score is useful when it helps us understand that experience. The number on its own cannot tell us whether the business has explained the offer or received a suitable enquiry.
- LCP
- When did the largest visible content element appear?
- INP
- How promptly did an interaction receive a visual response?
- CLS
- How much unexpected layout movement was measured?
These describe loading, responsiveness and stability. TBT is a separate lab diagnostic; a lab score is not a real-user average.
Start with the task, then the test.
I want to know what the visitor needs next. A builder might need project photos and a clear service area. A supplier might need a product comparison. Testing should include those paths, with the content and interactions people actually use.
A lean homepage is a useful starting point, but it does not prove the product gallery, checkout or enquiry form works well. Choose representative pages and record the device, connection, date and test conditions. Those details make later comparisons more useful.
Three different parts of the experience.
Largest Contentful Paint, or LCP, records when the largest visible content element is painted. Interaction to Next Paint, or INP, concerns how promptly the page responds to interaction. Cumulative Layout Shift, or CLS, measures unexpected movement. Each describes a different kind of friction.
A page can look ready while still responding poorly to a tap. It can also load quickly and move a button as an image arrives. Looking at all three helps identify the actual problem instead of treating speed as one finish line.
Reading behind this section: Philip Walton / web.dev
A lab result and real use answer different questions.
A lab test runs under a particular setup. Field data comes from recorded real use and reflects a wider mix of visits. A strong saved lab result is worth showing with its date and conditions; it is not an average for every customer.
Total Blocking Time, or TBT, is a lab measure of blocking during the measured period. A low TBT is useful diagnostic evidence, but it is not a replacement for field INP. Neither a test score nor a timing proves a sale.
Reading behind this section: Philip Walton and Barry Pollard / web.dev
Keep checking as the website changes.
New photos, tracking scripts, fonts and embedded tools can change the experience after launch. Agree a sensible performance budget and review the pages when important content or functionality changes. Harry Roberts describes budgets as safeguards against regression.
The business decision is which improvement removes useful friction. Keep the measurement beside a real customer task, and check that the task still works after the change.
Reading behind this section: Harry Roberts / CSS Wizardry
Put it to use
A useful check for your business.
- Choose a representative customer task and page.
- Keep the test date, device and conditions beside the result.
- Separate lab evidence, real-use data and commercial outcomes.
The thinking behind this guide
Sources & further reading.
Our practical interpretation, with the references below for the specific principles and further detail.
- Web Vitals
Definitions of loading, responsiveness and visual stability; not a claim that a score predicts sales.
- Total Blocking Time
TBT is a lab diagnostic and is not a substitute for measuring INP in the field.
- Web Performance Budgets, Pragmatically
Performance budgets as safeguards against regression, using representative measurements.
