A good PageSpeed score but users still complain: which speed measurement should you trust?

How do lab data and field data differ?

PageSpeed Insights combines two types of measurement:

• Lab data: a test in a controlled environment with defined device and network settings, useful for diagnosing issues and repeating tests.
• Field data: measurements from real visits, reflecting differences in devices, connections and user behaviour.

The two results can therefore differ without either being wrong. A controlled test does not necessarily represent your audience's conditions and may miss issues that appear while scrolling or interacting with a page. [2]

Our recommendation at CROMBA is to give each measurement a clear question: what can we diagnose and fix, and what are users actually experiencing?

Read the report's scope before its score

PageSpeed Insights presents real-user measurements from the Chrome User Experience Report, or CrUX, covering the previous 28 days. Google stresses that a good lab result does not guarantee a good experience for actual users. [1]

Before comparing reports, check the URL, device category, test date and type of data shown. Comparing a desktop test with a mobile test can lead to development work that misses the customer's problem.

We suggest documenting complaints precisely: which page, on which device, and which step took too long? Was it content appearing, a menu opening or a contact form submitting?

These details turn “the website is slow” into a problem the team can test again.

Why do some pages have no real-user data?

A page needs enough samples to appear in CrUX. If those are unavailable, the tool may show origin-level data covering the site's pages instead of the specific page. If the origin also lacks enough samples, real-user experience data will not appear. [1]

Missing data therefore does not establish good or poor performance. Check the measurement scope before attributing a site-wide result to one page.

For a low-traffic website, start with practical tests of important journeys and consider whether direct measurement of visitor experience is appropriate for the project.

Prioritise fixes around the customer journey

At CROMBA, we suggest choosing pages that support important decisions: the service page receiving ad traffic, a product page, a quote request form and checkout steps where relevant.

For each issue, record:

• The page and step affected.
• The evidence showing the problem.
• Its potential impact on using the website.
• The proposed change and how it will be checked.

For example, a form-submission complaint requires testing field completion, submission and the confirmation message. The issue may concern request processing or unclear feedback, which needs investigation beyond a performance score.

For websites serving Arabic- and Turkish-speaking audiences, test important journeys in both editions. Differences in content, fonts and page elements deserve attention within the implementation scope.

To review alignment between the ad, message and form, see our guide to improving landing pages before increasing Google Ads spend. When tracking requests, our GA4 lead measurement guide helps distinguish a submit click from a successful enquiry.

What should acceptance criteria for performance work include?

Before work begins, agree the target pages, test conditions, problems to resolve and evidence that will demonstrate completion with your provider.

After implementation, repeat the agreed tests and confirm that forms, buttons and content still work. Then monitor field data, remembering that it covers a preceding period: changes do not immediately isolate the effect of a newly published fix. [1]

Evaluate more than an improved score. Is the task easier to complete? Has the specific complaint disappeared? Has the website retained its functionality?

If the review also covers accuracy and completeness, distinguish technical fixes from refreshing existing content or writing a new article: each addresses a different gap.

Discuss with CROMBA the performance issues affecting your customer journey and how to prioritise them using evidence and clear acceptance criteria.