Most WordPress performance work starts and ends with a PageSpeed Insights score. Someone runs a test, sees a number in orange or red, makes some changes and runs the test again until the number turns green. Then they move on. Real user monitoring fills that gap.
The trouble is that the score isn’t what your visitors experience. It’s a single simulated page load on a simulated mid-range phone over a simulated slow network, run from a Google data centre. Real visitors use a huge variety of devices, connections and locations. They arrive with or without cached files, scroll, tap and navigate between pages.
Real user monitoring, usually called RUM, measures performance from actual visits. It’s the difference between testing a car on a track and watching how it drives on real roads. Both are useful. Only one tells you what your customers live with.
Lab data versus field data
Lab data comes from controlled tests: Lighthouse, PageSpeed Insights’ lower section, WebPageTest and similar tools. It’s repeatable and great for debugging, because you can change one thing and see the effect.
Field data comes from real visitors. It’s messy and varies a lot, but it’s the truth about user experience. It’s also what Google uses for Core Web Vitals in search ranking.
The two often disagree. A site might score 95 in Lighthouse but fail Core Web Vitals in the field because most visitors are on older phones. Or it might score 60 in the lab but pass in the field because visitors are mostly on fast desktop connections.
When they disagree, field data wins for deciding whether you have a problem. Lab data is then the tool for diagnosing it.
Free sources of field data
You may already have more field data than you realise.
Chrome User Experience Report (CrUX)
Google collects anonymised performance data from Chrome users who have opted in. This is the CrUX dataset. If your site gets enough traffic, CrUX data appears:
- At the top of PageSpeed Insights, labelled “Discover what your real users are experiencing”
- In Search Console’s Core Web Vitals report
- In the CrUX dashboard and the CrUX API, for more detail and history
CrUX data is a rolling 28-day window, so changes take weeks to show. It only includes Chrome users, and only sites and pages with enough visits. Smaller sites often only get origin-level data, meaning the whole site averaged together, rather than per-page data.
Search Console
The Core Web Vitals report in Search Console groups your URLs by status for each metric on mobile and desktop. It’s useful for spotting which templates have problems, because it groups similar URLs together. When you fix something, you can use “Validate fix” to have Google re-check over 28 days.
Why you might want your own RUM
CrUX is free and authoritative, but it has gaps:
- Slow to update, so you can’t quickly see the effect of a change
- No data for low-traffic pages
- Limited detail about which element or interaction caused a problem
- No way to segment by your own dimensions, like logged-in users or specific campaigns
Collecting your own RUM data fills those gaps.
Setting up RUM on WordPress
The web-vitals library
Google’s open-source web-vitals JavaScript library measures LCP, INP, CLS and other metrics in the browser exactly as Chrome does. It’s tiny and you can send the results anywhere.
A simple approach sends metrics to Google Analytics 4 as events:
In production, self-host the library rather than loading it from a CDN, and enqueue it properly through your theme or a small plugin. The attribution build of the library also reports which element caused LCP or which interaction caused poor INP, which is extremely useful for debugging.
Using GA4 for analysis
Once the events are flowing, you can build explorations in GA4 that show metric distributions by page, device category and country. It takes some setup, and GA4 isn’t designed primarily for performance analysis, but it’s free and you probably already have it.
Dedicated RUM services
Several services specialise in RUM and make analysis much easier. Some are paid, some have free tiers. Cloudflare Web Analytics includes basic Core Web Vitals data at no cost if your site uses Cloudflare. Hosting providers and performance services increasingly include RUM dashboards too.
When choosing a tool, look for per-page breakdowns, device and country segmentation, attribution data showing which element or script caused issues, and reasonable data retention.
Privacy considerations
RUM collects data about visitors’ devices and behaviour. Performance data is usually not personal on its own, but if you send it to an analytics tool that uses cookies or identifiers, consent rules may apply. Many sites collect only aggregated, anonymous metrics, or include RUM under their analytics consent category.
Reading RUM data
Look at the 75th percentile
Google assesses Core Web Vitals at the 75th percentile. That means 75% of visits must be at or better than the threshold. Averages hide problems: a site with mostly fast visits and a long tail of very slow ones can have a decent average but a poor 75th percentile.
Most RUM tools let you view percentiles. Get used to looking at p75 rather than averages.
Segment by device
Mobile and desktop usually behave very differently. Analyse them separately. If your audience is mostly mobile, mobile p75 is the number that matters most.
Segment by page type
Group pages by template: homepage, blog posts, product pages, category pages, checkout. Problems are usually template-specific. If all product pages have poor LCP, the issue is in the product template, not in individual products.
Segment by country or connection
If you serve an international audience, performance in distant regions can be much worse. That’s a sign to look at CDN coverage or server location.
Turning data into action
RUM data is only valuable if it leads to changes. A practical routine:
- Each month, check p75 LCP, INP and CLS for mobile, by page template.
- Pick the template with the worst metric that affects the most traffic.
- Use attribution data to identify the cause: the LCP element, the slow interaction, the shifting element.
- Reproduce it in the lab with DevTools, using similar device throttling.
- Fix it, deploy and watch the RUM data over the following weeks.
Keep a simple log of changes and dates. When a metric improves or gets worse, you can match it to what changed.
Watch for regressions
One of the most valuable uses of RUM is spotting regressions. A plugin update adds a heavy script. A marketing team adds a new tag. A theme change breaks image sizing. Without monitoring, you find out weeks later from Search Console, or never.
Set up alerts if your RUM tool supports them, for example when p75 LCP on mobile rises above 3 seconds. Even a monthly review catches most problems before they become serious.
A lightweight setup for small sites
If a full RUM tool feels like too much, a minimal setup still gives you useful data. Enqueue the web-vitals library with a short script that sends LCP, INP and CLS to GA4, as shown above, and create one exploration in GA4 that shows the 75th percentile of each metric by page path and device category. Review it once a month alongside your other analytics.
That costs nothing, takes about an hour to set up and gives you per-page field data that CrUX may not provide for a smaller site. It’s also a good way to prove the value of performance work to clients or managers, because you can show real visitors’ experience improving rather than a lab score changing.
Sharing performance data with the team
Performance improves fastest when the people making changes can see the effect. Share a simple monthly summary with whoever edits the site: the three Core Web Vitals for mobile, a comparison with the previous month and a note about any changes that likely caused movement. When content editors learn that a new 4 MB banner image made LCP worse, they tend to compress the next one. When developers see INP improve after removing a script, they’re more willing to question the next script someone wants to add.
Frequently asked questions
Why does Search Console say my pages are slow when PageSpeed shows a high score?
Search Console uses field data from real users, while the PageSpeed score is a lab test. Real visitors may have slower devices or networks, or interact in ways the lab test doesn’t.
My site doesn’t have CrUX data. What should I do?
Smaller sites often don’t have enough Chrome traffic for CrUX. Collect your own RUM with the web-vitals library or a RUM tool, and rely on lab testing on realistic devices.
Does adding RUM slow my site down?
The web-vitals library is very small and runs after the main content. When sent efficiently, its impact is negligible.
How long until improvements show in Google’s data?
Up to 28 days for CrUX and Search Console. Your own RUM shows changes within days, as soon as enough new visits are recorded.




