PageSpeed testing: why it matters and how to check your website

A visitor opens your website to check a price, book a service or buy a product. If the main content appears late, the menu responds slowly or a button moves just before a tap, the website is making that task harder.

A PageSpeed test helps you identify performance problems and decide what to improve. It is useful before launch, after significant changes and throughout the life of the website. At B2BWebs, regular checks are also part of ongoing care: client websites built by us and covered by our monthly hosting and support service receive weekly LimeSeek checks and automated reports, at no additional charge.

This guide explains how to run your own test, read the results and use them sensibly. The technical references were checked on 28 October 2026.

Why website performance matters

Performance affects the practical experience of using your site. A customer should be able to see what you offer, navigate between pages and complete an enquiry without unnecessary waiting or unexpected movement.

Think about the pages that lead to business results: a product, a service description, a booking page or the contact form. A fast homepage is useful, but it does not tell you whether the rest of that journey works well.

Performance also matters to search visibility. Google uses Core Web Vitals in its ranking systems, while making clear that good scores do not guarantee a leading position. Useful content, relevance and the overall experience still matter. Google: page experience in Search.

What PageSpeed Insights actually measures

Google PageSpeed Insights offers a browser-based way to examine a page on mobile and desktop. Its report brings together two kinds of evidence:

• Real-user data: recent visitor experience from Chrome's CrUX dataset, when enough data is available. This covers a rolling 28-day period and can describe either the specific URL or the whole origin.
• Lab data: a Lighthouse test in simulated conditions, useful for diagnosing the current page and finding possible improvements.

A small or new website may have no field data. That is a lack of measurements, not proof that the website is slow. Check which data level the report shows. Google: how PageSpeed Insights works.

Three measurements worth understanding

Core Web Vitals describe different parts of the experience:

Metric	What it tells you	Good target
LCP — Largest Contentful Paint	How quickly the largest visible content element appears	2.5 seconds or less
INP — Interaction to Next Paint	How quickly the page visibly responds to interaction	200 milliseconds or less
CLS — Cumulative Layout Shift	How much the layout shifts unexpectedly	0.1 or less

These are experience measurements, not three versions of the same speed score. Google: Core Web Vitals.

The good thresholds apply at the 75th percentile of recorded visits, assessed separately for mobile and desktop. A standard Lighthouse loading test does not measure INP; its Total Blocking Time (TBT) can help diagnose responsiveness, but does not replace real-user INP. Google web.dev: measuring Web Vitals.

The Lighthouse Performance score is a separate summary on a 0–100 scale: 90–100 is good, 50–89 needs improvement and 0–49 is poor. A score of 100 does not mean that a page takes zero time to load, and a green score does not certify every function of your website. Chrome: Lighthouse scoring.

How to test your website, step by step

1. Open PageSpeed Insights. Paste the full address of a public page, including https://, and select Analyze. For example, test the actual product or service page your customers visit.
2. Review Mobile and Desktop separately. Start with the mobile report, then switch to desktop. Save which mode you tested; the results represent different test conditions.
3. Check whether visitor data is available. Read the real-user section before drawing conclusions from the lab score. Note whether it refers to this URL or the whole origin.
4. Read the explanations below the score. Look for the resources, images or scripts named in the diagnostics. Send the report link and affected URL to the person responsible for the site, rather than sending only a screenshot of the score.
5. Repeat for representative pages. Our recommended starting set is the homepage, an important service or category page, a product or booking page, and the contact page. Add a manual check of the enquiry or purchase flow on a real phone.
6. Keep a record and retest after changes. Save the date, page, device mode, important metrics and the change you made. Compare the same page and mode, keeping the test conditions as consistent as possible.

For an important decision, we recommend several runs rather than choosing the best result. Scores can vary as underlying conditions change, so a small movement in one run is not enough to prove a regression or an improvement. Chrome: understanding score variation.

What to improve first

Use the report to identify a cause before changing the site. Common starting points include:

• Main content that appears late: inspect the server response, the main image and resources that delay rendering. An image visible immediately on opening the page should not be delayed with unnecessary lazy loading. Google web.dev: improving LCP.
• Controls that feel unresponsive: inspect long-running JavaScript and the work performed when someone clicks, taps or types. Removing unnecessary work can be more useful than changing the visual design. Google web.dev: improving INP.
• Content that jumps around: reserve space for images and embedded content, and review how fonts and late-loading elements affect the layout. Google web.dev: improving CLS.

Prioritise issues on pages customers depend on. After a change, check that menus, forms, consent controls and purchasing still work. A performance improvement should make the website easier to use.

Why one test at launch is not enough

A website changes over time. A new banner, video, external widget, tracking script or redesigned section can change what the browser has to load. A result saved on launch day is a useful baseline, but it does not describe every later version of the site.

Regular testing makes it easier to notice changes and decide when a closer investigation is needed. It also provides a history to refer to after updates, instead of relying on memory or waiting for a customer to report that a page feels slow.

For a practical example of what to investigate, read what to check first on a slow mobile website.

Weekly LimeSeek checks are included with B2BWebs hosting and support

Websites built by B2BWebs and maintained under our monthly hosting and support service are tested every week through LimeSeek. Automated reports are included in that service, with no additional charge for the weekly checks and reports.

This gives performance monitoring a regular place in the maintenance process. The reports provide recorded findings that help us compare checks, identify changes and decide where follow-up work is needed. Clients do not have to remember to organise a new manual test each week.

LimeSeek brings technical website checks and performance results into a reporting workflow. The coverage and available measurements shown in each report matter: a check of selected pages should be read as evidence about those pages, rather than a guarantee about every page or every visitor.

The purpose is to keep watch over the website as it evolves and make problems easier to investigate. An automated report helps identify what needs attention; it does not mean that every issue is automatically repaired or that a website will always score 100.

Learn more about B2BWebs and LimeSeek and what website maintenance involves.

Start with the pages that matter to your customers

Run a mobile test on one important page, read the findings and check the same journey on your phone. If something needs attention, keep the report and use it to agree the next improvement.

If you would like help interpreting a result or discussing ongoing website care, contact B2BWebs. For websites we build and cover through our monthly hosting and support, weekly LimeSeek testing and automated reporting are already part of the service.