Mobile page speed: how to read your PageSpeed score
Got a red PageSpeed Insights score and not sure what to do next? How to read the report and focus on the signals that actually matter for mobile visitors.
You type your website's address into PageSpeed Insights and a red circle appears. The desktop result looks fine, but the mobile score is low. Below it comes a long list of suggestions: "reduce unused JavaScript", "serve images in next-gen formats", "eliminate render-blocking resources". Which of these actually matter?
In this post we want to turn the PageSpeed report from a source of anxiety into a roadmap. The goal is not a perfect 100; it is a better experience for people on their phones. We look at speed the same way we look at everything else, through WIFT: "What's In It For Them?" What does this improvement actually give the person on the other side of the screen?
The report has two faces: real users and a lab test
PageSpeed Insights does not show a single measurement. It puts two different data sources side by side. According to Google's own documentation, the top section is field data collected from real Chrome users and summarises the previous 28 days. The section below it is a lab test run by a tool called Lighthouse, using one simulated device and one network profile.
The distinction matters because the two answer different questions:
- Field data answers "How do my real visitors experience the site?" This is what Google relies on when it assesses page experience.
- Lab data answers "What technical problems does this page have, and where should I start?" It is repeatable, so it is useful for checking a change.
On sites with modest traffic, field data may not appear at all. In that case the lab results are your guide, as long as you remember they come from a single simulation.
The big coloured number: the Performance score
The 0-100 score in the middle comes from the lab test. The same documentation classifies 90 and above as "good", 50-89 as "needs improvement" and anything below 50 as "poor".
A few things are worth knowing about that number:
- It moves between runs. Server load, third-party scripts and network conditions shift the result, so test a few times and look at the trend.
- The mobile test is deliberately demanding. Lighthouse's mobile run emulates a mid-range phone on a slow mobile connection, so a mobile score lower than the desktop one is normal.
- The score itself is not a ranking factor. For search visibility, what counts is the Core Web Vitals data from real users. The score is a tool that helps you improve those values.
What to look at first: Core Web Vitals
Core Web Vitals are the three metrics Google uses to summarise user experience. Based on the definitions on web.dev, each has a "good" threshold, and a page is assessed at the 75th percentile of visits:
- LCP (Largest Contentful Paint): how long it takes for the largest image or text block to appear. Good is 2.5 seconds or less.
- INP (Interaction to Next Paint): how quickly the page responds when you tap a button or open a menu. Good is 200 milliseconds or less. It answers "does the site feel frozen?"
- CLS (Cumulative Layout Shift): how much the content jumps around while loading. Good is 0.1 or less. Think of reaching for a link just as an image loads above it and pushes everything down.
At the top of the report, each of these is marked green, orange or red. When prioritising, fixing the metric that is red or orange is far more meaningful than nudging the score up a few points.
How to read the list of suggestions
The diagnostics and opportunities at the bottom of the report can feel overwhelming. In our experience, this order works well:
- Identify which metric is struggling. If LCP is slow, look at images and server response time. If INP is slow, look at JavaScript. If CLS is high, look at images without set dimensions, ad slots and late-loading fonts.
- Connect each suggestion to that metric. Lighthouse lets you filter suggestions by the metric they affect. Anything that does not touch your problem metric can wait.
- Pick the biggest gain for the least effort. Serving the main hero image at the right size and compression often makes a visible difference on its own, while removing a plugin takes more work.
Common mobile issues we see
Every site is different, but certain issues come up again and again in the sites we review:
- Large, uncompressed hero images. Sending a wide desktop image unchanged to a phone directly lengthens LCP. Responsive sizes and modern formats such as WebP or AVIF reduce that weight.
- Too many third-party scripts. Chat widgets, analytics tools, tracking pixels and social embeds each look small, but together they keep the phone's processor busy and worsen INP. Start by asking whether each one is still needed.
- Images and slots without reserved space. If the height of an image, video or ad slot is not reserved in advance, the page shifts as content arrives and CLS goes up.
- Heavy sliders and animations. Large auto-rotating sliders at the top of the page can strain both LCP and responsiveness. A single strong image often delivers the same message faster.
- Slow server response. If a lot of time passes before the page even begins to render, front-end improvements can only do so much. Hosting, caching and CDNs matter here.
Measure experience, not just the score
Some changes made purely to reach 100 can remove parts of the site that actually do a job: delaying the contact form, over-compressing product photos, or stripping out measurement tools. So for every change it helps to ask: does this make things easier for the visitor on their phone?
After a change, you will see the lab result improve straight away. Field data, however, summarises a 28-day window, so it can take a few weeks for the real-user impact to show up in the report. The Core Web Vitals report in Google Search Console helps track this across groups of pages.
In short
Read correctly, the PageSpeed report is a valuable tool that shows exactly where your site struggles on mobile. Look at real-user data and Core Web Vitals first, then prioritise the suggestions that affect your weakest metric, and treat the score as a progress indicator rather than the goal.
If you would like a second pair of eyes on your site's mobile report, we would be happy to take a look together.