Is My Website Too Slow on Phones?

You checked your site on your own phone and it seemed fine. Then PageSpeed Insights handed you a red number, three acronyms and a list of fixes nobody explained. A wrong guess costs you either way: a rebuild your visitors never notice, or months of paid mobile clicks that leave before the page appears.
A website is too slow on phones when its mobile field data fails Core Web Vitals, meaning real visits read at the 75th percentile against Google's three thresholds. A red lab score and your own phone are diagnostics, not verdicts. Fix first the pages your ad budget lands on, where every visitor was paid for before the page loaded.
What do Core Web Vitals actually measure?
Core Web Vitals are Google's three measurements of how a page feels to a real visitor. Each one maps to a moment where someone on a phone gives up.
- Largest Contentful Paint (LCP) times how long the main content takes to appear. That is usually the hero image or the headline block.
- Interaction to Next Paint (INP) measures the gap between a tap and the screen visibly responding, across every interaction in the visit.
- Cumulative Layout Shift (CLS) scores how much the layout jumps while it loads. It is the reason you sometimes hit the wrong link.
The vocabulary matters because it turns "slow" into three separate problems, usually with three separate owners:
- A poor LCP tends to point at heavy images or a slow server.
- A poor INP points at too much JavaScript tying up the browser. On marketing sites, much of that JavaScript is the marketing team's own tags.
- A poor CLS points at images, banners, ads or fonts arriving without space reserved for them.
Hand the wrong metric to the wrong person and nothing improves, however much effort goes in.
Two details catch out most first-time readers. First, INP replaced First Input Delay (FID) in March 2024. FID only timed the wait before the first interaction began to be processed. INP watches every interaction through to the moment the screen updates, which makes it harder to pass and a fairer picture of using the page. Any audit or agency deck that still lists FID predates the switch, and its advice on responsiveness is incomplete.
Second, scores are assessed page by page and device by device. Your homepage can pass on desktop while your main landing page fails on mobile. "My site" is rarely one answer.
On rankings, Core Web Vitals are one page experience signal among many. Google recommends getting them into good shape but has never published how much weight they carry. Relevance still beats speed: a fast page with the wrong answer will not outrank a slower page with the right one.
The stronger reason to care is commercial. The blank hero, the button that does nothing and the jumping layout cost you visitors whether Google is watching or not. On paid traffic, you have already bought the click. The full explanation of each metric, and why FID disappeared, is in what Core Web Vitals are, in plain English.
What score counts as fast enough on mobile?
This is where "too slow" gets an exact meaning. Google has used these bands since INP joined the set in March 2024:
| Metric | Good | Needs improvement | Poor | | --- | --- | --- | --- | | LCP | 2.5 s or less | 2.5 s to 4 s | Over 4 s | | INP | 200 ms or less | 200 ms to 500 ms | Over 500 ms | | CLS | 0.1 or less | 0.1 to 0.25 | Over 0.25 |
The thresholds are the easy part. How they are applied decides your verdict. Google reads each metric at the 75th percentile of real Chrome visits over a rolling 28 days, so at least three in four visits have to clear the line.
That choice is deliberate. The median would let a page pass while half its visitors had a poor experience. The 95th percentile would hand the result to a few old phones on terrible connections, whatever you did to the page. The 75th sits between the two. It is strict enough to cover most of your audience and stable enough that you can actually move it.
Three consequences follow:
- The verdict is all-or-nothing. A page with good LCP and CLS but a 260 ms INP fails the assessment. Treat amber as a failure you have a head start on.
- CLS is a unitless score, not a time. A CLS of 0.1 has nothing to do with a tenth of a second.
- A threshold is a floor, not a target. A page sitting at 2.4 s LCP is one heavier hero image or one new script away from amber next month.
This is also why the question is specifically about phones. The same page, with identical code, runs on slower processors and patchier networks on mobile. The slow tail of visits stretches further, so the 75th percentile lands further out. Green on desktop and amber on mobile is the normal pattern, not a sign something is broken. For most sites, mobile is where the assessment fails and where most visits come from, so it is the score to read first.
These bands are Google's to change. Confirm them against PageSpeed Insights or Search Console, which color their results using the same thresholds. The full working on percentiles, the 28-day window and which metric to fix first is in what a good LCP, INP and CLS score is.
Which number should you trust in PageSpeed?
Most "is my site slow?" arguments come from reading two different measurements as one. PageSpeed Insights shows both on the same page.
Field data sits at the top. It shows how real Chrome users experienced the page over the last 28 days, at the 75th percentile, drawn from the Chrome User Experience Report. That is the Core Web Vitals assessment, and it is the version that counts.
The big colored score below it is a Lighthouse lab test. Lighthouse runs one simulated load on an emulated mid-range phone over a throttled connection, with nothing cached. That makes it repeatable and good for diagnosis. But it describes one imagined visit, not your audience.
Your own impression is the weakest of the three readings. You are probably on a recent phone, on Wi-Fi, on a site you have opened dozens of times. Every image and font is cached, and your thumb already knows where to tap. That is the best-case visit, and few of your visitors get it. So when the lab says slow and the site feels fast, it usually means the page is heavier than it needs to be and your hardware hides the weight.
Let the field section decide:
- Field data passes at page level. A red lab score is a backlog, not an emergency. Work through the cheap diagnostics and do not let it drive a rebuild.
- Field data fails. Act on it, however fast the site feels to you, and use the lab diagnostics to find the cause.
- There is no page-level field data. The lab is your best evidence. New and low-traffic landing pages often fall back to origin-level data, which describes the domain average, not the page in front of you.
One combination deserves more worry than a red lab score: poor field data alongside a clean lab run. Lighthouse loads the page once and clicks nothing. It cannot measure INP at all, so Total Blocking Time stands in as a proxy. It also misses chat widgets, consent tools and scripts that only fire on interaction. If real users struggle and the lab does not, one of two things is true. The problem sits in something the simulation never does, or your audience is on slower connections than you assumed.
There is an apparent contradiction across this cluster worth resolving. One piece tells you to load the page on your own phone, and another calls your phone the least reliable reading. Both hold.
A phone test is only useful once you strip away your advantages. Use a mid-range device, mobile data and a cold cache, on the page visitors actually land on. Done that way, it is a sanity check on the field data. Done on office Wi-Fi, it is flattery. The full breakdown is in PageSpeed says my site is slow but it feels fast.
Does a slow site raise your Google Ads costs?
Yes, through two routes, and the one most people worry about is usually the smaller.
The direct route: Quality Score
Google rates each keyword from one to 10 on three components: expected clickthrough rate, ad relevance and landing page experience. Google's guidance on landing page experience lists load time alongside relevant content, trustworthiness and easy navigation. Weaker quality means bidding more to hold the same position. A bad enough landing page experience can drop the ad below the threshold it needs to show at all.
Nobody outside Google can tell you how many cents a second of load time adds to a click. Google does not publish the weighting, and agency posts claiming slow pages "inflate" cost per click do not produce a measured figure. Be suspicious of any precise number you are quoted.
Landing page experience is also only one input of three. A fast page that fails to deliver what the ad promised loses more quality than a sluggish page that delivers it.
The indirect route: clicks that never convert
This route is easier to see and usually larger. Google charges for the click the moment it happens, whether or not your page ever finishes loading.
Suppose you pay $2 a click and one in 20 visitors buys, so each sale costs $40 in ad spend. If a slow mobile page drops that to one in 25, the same clicks at the same price cost $50 a sale. Cost per click never moved, but cost per acquisition rose 25%. Those figures are an illustration, not a benchmark. The mechanism is why a speed problem so often shows up in the account looking like a bidding problem.
Two systems, not one
The cluster draws a distinction worth holding onto. Core Web Vitals belong to Google's search tools, while landing page experience is Google Ads' own assessment. Google does not publish that the ads auction reads your Core Web Vitals scores directly. Google Ads also publishes no load-time cut-off.
The two measure overlapping things, and a page that loads quickly on a mid-range phone tends to do well on both. But passing one does not guarantee the other. For an advertiser, the more telling number is often mobile conversion rate against desktop on the highest-spending landing page.
When speed matters most
Speed matters most when clicks come from phones and the searcher wants something now. Think of someone finding an emergency plumber on mobile data, or checking prices while standing in a shop. It matters less when traffic is mostly desktop broadband, landing page experience already reads "Average" or better, and conversion is healthy. The full diagnosis, including what to check before you raise bids, is in does a slow website make my Google Ads cost more.
What usually makes a small site slow?
Small business sites rarely get slow from one bad decision. They get slow from a pile of reasonable ones, and the culprits usually come down to four. Each leaves a distinct fingerprint in a speed report, and that fingerprint connects the metrics above to a fix.
Oversized images
Someone uploads a photo straight from a phone camera or a stock library at full resolution. The site then displays it in a box a few hundred pixels wide, and the visitor downloads every pixel anyway. When PageSpeed names an image as the largest element on screen, this is your LCP problem.
- Resize each image to the size it is actually displayed at.
- Compress it, and serve WebP where your platform supports it.
- Lazy-load images below the fold.
Keep doing it, because every full-size upload brings the problem back.
Marketing tags
This is the cause marketers own, and it accumulates one campaign at a time. A conversion tag goes in, a heatmap trial is never removed, a LinkedIn pixel arrives, then a review widget. Analytics ends up installed twice, once by a plugin and once by a tag manager.
Each script runs from another company's server and can pull in more scripts from yet more domains. Each one competes with the page for the browser's main thread, which is the INP problem. Vendors also change their scripts whenever they like, which is how a site that was fine in January gets slower by June. The fingerprint is long blocking time from other companies' domains.
Hosting and caching
If the page sits blank before anything appears, the problem is upstream of images and scripts: the server is slow to send its first byte. The usual causes are cheap shared hosting, no content delivery network, and redirect chains that add round trips before the real page starts loading.
Missing page caching is the most common of all. Without it, WordPress rebuilds each page from the database on every visit, which falls over at exactly the moment a campaign is paying to send traffic. Turn caching on first. Move to better hosting only if server response is still slow with caching in place.
Themes and plugins
Multipurpose themes ship code, fonts and effects for every demo layout they advertise, and your site loads much of it regardless. Plugins outlive the redesigns that needed them. Quality matters more than count: one badly built plugin can cost more than ten good ones.
The test is simple. Install the theme on a bare site and run a speed test. If it is slow there, tidying plugins will not rescue it.
Where to start
Order matters. Images and redundant tags cost least to fix and rarely break anything. Hosting moves and theme changes are real projects, worth taking on once you know they are the bottleneck.
Run PageSpeed on the mobile view of the page your ads point at, and read it by symptom rather than score. Then stop the problem coming back: give every new tag a named owner and a removal date, and recheck speed after each launch. The full list, with how to tell each cause apart, is in what usually makes a small business website slow.
Conclusion
Whether your site is too slow on phones is a question with a precise answer, not a feeling. Read the mobile field data, page by page, and compare it against LCP of 2.5 seconds, INP of 200 milliseconds and CLS of 0.1 for three in four real visits. The lab score tells you why a page is slow. Your own phone tells you almost nothing unless you deliberately make it a worse phone.
When a page fails, the cost is rarely a dramatic ranking drop. It is paid mobile clicks that leave before the page appears, and a cost per sale that climbs while cost per click looks fine. The causes accumulated in small pieces, and they come apart the same way. Fix images and redundant tags first, and fix the pages carrying your budget before the ones nobody reaches.
To see which landing pages carry your paid traffic and how their mobile conversion rate compares with desktop, SproutMe Companion answers that from your connected Google Ads and Analytics data.
Frequently Asked Questions
Longer than most people wait. Field data covers a rolling 28-day window, so a fix shipped today is diluted by weeks of older, slower visits. It shows fully only after about four weeks. Google Ads also reassesses landing page experience over time. Mobile conversion rate usually responds fastest, and a lab rerun confirms the same day that the fix behaves as intended.
Only part of it. Page caching stops WordPress rebuilding each page from the database on every visit, which cuts server response time and helps most under heavy traffic. It does nothing about oversized images, third-party scripts or a bloated theme, because those load in the visitor's browser after the server has replied. If the slowness is in the browser, caching barely moves it.
The container itself is light, but it makes adding tags so easy that they pile up without review. Every tag it fires still downloads and runs its own code. It can also duplicate tracking a plugin already installed, which slows the page and double-counts visits. Audit what it fires regularly, and remove anything without an owner or a current campaign behind it.
Get a complimentary audit to uncover AI opportunities hidden in your data.
Put these strategies to work


