Your page moves while it loads after the browser gets pieces late and no room was saved for them. An image with no set size, a font swap, a cookie bar or a chat box drops in above your phone number, and everything below it slides down. The tap lands on the wrong thing.
The jump has a name and a score
Google calls it Cumulative Layout Shift, or CLS. Its own web.dev page, last updated April 12, 2023, defines it as “a measure of the largest burst of layout shift scores for every unexpected layout shift that occurs during the entire lifecycle of a page.”
The target is 0.1 or less, measured at the 75th percentile of page loads, split between mobile and desktop. Three visits out of four have to come in at 0.1 or under. Your best visit does not count. Your worst quarter does.
That same page is blunt about why it matters. Unexpected shifts disrupt the visit in many ways, it says, “from causing them to lose their place while reading if the text moves suddenly, to making them click the wrong link or button.” One of its own examples: “A sudden shift in layout makes the user confirm a large order they intended to cancel.”
Not every move is a fault. web.dev draws the line simply: “A layout shift is only bad if the user isn’t expecting it.” A menu that opens after someone taps the menu button is expected. A phone number that slides two inches down half a second after the page appears is not.
Why the phone number is the part that gets hit
Your number sits near the top. Anything that arrives late and lands above it pushes it down.
Think about how the visit really goes. A water heater is leaking. The customer is standing in a garage with one bar of signal. Your page paints, they see the number, the thumb is already moving. Then the logo image finally arrives, claims its real height, and the whole page steps down. The thumb lands on the service area link instead.
They do not think “layout shift”. They think the site is broken, and they hit back.
That is why a jump costs a trades site more than it costs a blog. A reader who loses their place scrolls back up. A person with water on the floor picks the next name on the list.
Four things push your page around
The web.dev guide to fixing this, last updated February 7, 2025 and credited to Addy Osmani and Barry Pollard, names the causes in this order:
- Images without dimensions
- Ads, embeds and iframes without dimensions
- Content dropped into the page after it loads, again with no size set
- Web fonts
The pattern is the same every time. Something shows up late, and nothing was holding a seat for it.
Images are the big one, and the trouble is not the file size. It is the missing width and height. The browser cannot save the right amount of space for a picture it has not measured yet. The HTTP Archive Web Almanac 2025 performance chapter, published January 15, 2026 on data collected in July 2025, found 62% of mobile pages fail to set dimensions on at least one image. On desktop it is 65%.
Sit with that number for a second. Most of the web has this fault. It is common, and it is cheap to fix. File size is a separate problem, and oversized jobsite photos drag a page down a different way.
Fonts cause a smaller version of the same jump. Text draws in a backup font, the real font lands, and the lines re-flow. The same Almanac chapter found 87% of mobile pages use at least one web font. There is more on what a custom font costs you if that is where your jump comes from.
The score can look clean while customers still get moved
Here is the part a speed report will not tell you.
The CLS page says shifts that happen within 500 milliseconds of user input get a flag set on them, “so they can be excluded from calculations.” So if your page moves right after somebody taps, that move may never reach your number, even though the customer felt it.
Lab tests miss more than that. Google’s own page on why lab and field data differ, last updated July 18, 2022 and written by Philip Walton, says “CLS measured in the lab only considers layout shifts that occur above the fold and during load, but this is only a subset of what CLS actually measures.” Out in the real world, it says, the score counts shifts across the whole life of the page, including “content that shifts as the user scrolls.”
So a clean test score is not proof your page holds still. It is one load, on one connection, with nobody scrolling and nobody tapping. If your report shows no real visitor data at all, that is its own separate problem.
When you should not spend money on this
If your page already comes in under 0.1, stop reading and go do something else. Do not pay anyone to fix a jump you do not have. The Web Almanac found 81% of mobile pages already score good on this measure, up two points from 2024. The odds are decent that yours is one of them.
Fixing the jump is also not a ranking play. Google Search Central’s page experience document, last updated December 10, 2025, says “Core Web Vitals are used by our ranking systems.” The very same page then warns that good results in Search Console or third party tools “doesn’t guarantee that your pages will rank at the top of Google Search results.” Both sentences are Google’s, and both are true at once. Read them together, and do not let anybody quote you only the first one.
And none of this helps if the number on the page is out of date, the hours are wrong, or nobody picks up the phone.
What to do next
- Load your own site on your phone, on cell data, not on office wifi. Watch the top of the page for the first three seconds. If your number moves, you found it.
- Run Google PageSpeed Insights on your home page and on one service page. Read the mobile CLS number. Write it down with the date next to it.
- Ask whoever built the site to put a width and a height on every image, starting with the logo and the top photo.
- Hold space for anything that lands late. A cookie bar, a review badge, a chat box. The gap has to be reserved before the thing arrives, not made when it drops.
- Check again in a week. Same page, same test, same tab.
Most of the time that list is short and the site you have can be fixed in place. If the site runs on a page builder with a stack of plugins, the list gets long, and there is a point where a plain rebuild costs less than the patching, which is its own decision to work out on paper. That point is real. It also sits later than most people selling rebuilds will admit.
The only proof this site owns is its own build: it scored 100 on Google’s PageSpeed test, desktop and mobile, when it launched in August 2026. That is one site, this one, and it is not a promise about yours.