PageSpeed Insights Score Low on Mobile but Fine on Desktop

Your PageSpeed Insights score is low on mobile — often red or orange — while desktop looks fine, and you want to know whether the mobile number is real, why it jumps around, and which fixes actually raise it.

Common signs of this issue

• PageSpeed Insights shows a low score on mobile — say 30 to 60 — while the desktop tab scores 80 or higher.
• The mobile score changes by several points each time you run the test, with no changes to the site.
• The report lists items like "Reduce unused JavaScript", "Eliminate render-blocking resources", or "Largest Contentful Paint element".
• Total Blocking Time or Largest Contentful Paint is shown in red on mobile.
• The site feels reasonably quick on your own phone over Wi-Fi, so the score seems unfair.
• An agency, SEO tool, or cold email pointed at your mobile score as proof the site is broken.

Safe checks you can do yourself

None of these require sharing passwords with anyone.

• Run the page at pagespeed.web.dev and look at the top section first. If it shows a Core Web Vitals Assessment from real users, that field data matters more than the colored score. If it says there is not enough real-user data, which is common for small sites, the lab score is all you have — useful for diagnosis, but still a simulation.
• Run the test three times and note the middle score. A swing of five to ten points between runs is normal; don't celebrate or panic over one run.
• Look at the five metrics under the score. On mobile, Total Blocking Time and Largest Contentful Paint usually carry most of the weight, so whichever of those is red is where the points are going.
• Find the LCP element in the diagnostics — usually your hero image or a large heading. If it is a big photo, check its file size and dimensions; a 3 MB, 4000-pixel image on a phone screen is a common culprit. See how to reduce image size.
• Check whether that main image is lazy-loaded. Lazy loading is good for images further down, but on the first image people see it delays the most important paint. Many themes and plugins let you exclude the first image.
• Read the third-party section, often labelled something like third-party code or third parties. Chat widgets, embedded video, maps, review widgets, booking tools, and multiple tracking pixels are frequently the biggest mobile drag. List them and ask which ones actually earn their place.
• Open Reduce unused JavaScript and render-blocking items and note which plugin or service each file belongs to (the file path usually names it). One page builder or slider loading everywhere often explains most of it. See too many plugins slowing the site.
• Compare a simple page, like your privacy policy, with your homepage. If the simple page scores well, the problem is what is on the homepage. If both are poor, look at hosting, caching, and site-wide scripts.
• Finally, load the site on an older phone over mobile data, not Wi-Fi. That is roughly what the mobile test imitates, and it tells you whether real customers feel the slowness.

What this usually means

The mobile score is lower because the test is deliberately harsh. PageSpeed Insights runs Lighthouse with simulated throttling: it imitates a mid-range phone with a slower processor and a slow mobile connection. Your desktop test gets a fast machine and fast connection. The same page, same code, simply takes much longer under those conditions, and JavaScript suffers most because a slower processor takes longer to run it. That is why Total Blocking Time — how long the page is too busy to respond — is so often the mobile problem, and why sites heavy on page builders, sliders, and third-party widgets drop hard on mobile while looking fine on desktop.

Keep lab and field data separate in your head. The 0 to 100 score is lab data: one simulated load, useful for spotting problems, and it varies between runs. Field data — the Core Web Vitals assessment, when your site has enough Chrome visitors to show it — is what real people experienced over the past 28 days, and it is what Google uses in search. A mobile lab score of 55 with a passing field assessment usually means real visitors are fine. For the metrics themselves and what counts as good, see Core Web Vitals failing.

A handful of items account for most mobile points: a large or lazy-loaded LCP image, render-blocking CSS and scripts in the page head, unused JavaScript from plugins that load on every page whether needed or not, and third-party scripts you added over the years. Fixing those usually moves a mobile score from red into orange or green. Chasing 100 is a different matter — the last points often mean removing features customers use, like chat or embedded reviews, and a perfect lab score is not a ranking factor. Aim for a site that feels fast on a real phone and passing field data where available. If the whole site is slow, not just the score, start with website loads slowly.

What not to do

• Don't chase a perfect 100 on mobile. Past a reasonable score, the gains are tiny and often cost features your customers use.
• Don't trust a single test run. Scores vary between runs, so compare the middle of three.
• Don't switch on every optimization option in a speed plugin at once. Delaying or combining all scripts is a common way to break menus, forms, and checkout.
• Don't lazy-load the first large image on the page. It usually makes the mobile score worse, not better.
• Don't pay for a monthly speed service based on a single red screenshot. Most sites need a one-time cleanup, not a subscription.
• Don't remove tracking, chat, or booking tools without checking who relies on them. Ask first, then decide.

When to get help

It is worth bringing someone in when the mobile score is poor and the site also feels slow on a real phone, when the report points at theme or page-builder files you can't safely change, or when every speed plugin you try breaks something. The biggest improvements often come from decisions a report can't make for you — which scripts to load only on the pages that need them, whether a heavy theme is worth keeping, whether the host is the bottleneck — and from testing each change so forms and checkout keep working. That judgment is what separates a real improvement from a score that moves while the site breaks.

Glenn, who runs WebsiteSelfHelp, improves mobile speed on small business sites without breaking what already works. Send the page you are worried about and you will get a straight answer on what is slowing it down on phones and what fixing it would take. No passwords needed to start.

Frequently asked questions

Related free guides