44% of the Web Just Flunked Google's Speed Test on Real Phones, and the Report That Used to Tell You Is Gone
You ran the speed test. It came back green. You closed the tab and went back to running your business.
Here's the part nobody mentioned: that green score isn't the one Google ranks you on.
The number that actually decides your search tiebreaker is measured on your real customers' phones, over the last 28 days, and most small business owners can't even see it anymore. Let me walk you through what's going on, because it's more fixable than it sounds once you understand what's being measured.
The 44% nobody's talking about
Google publishes real-world performance data every month from actual Chrome users. The batch from May 2026, released June 9, looked at about 18.4 million websites. Only 55.9% passed all three of Google's speed and stability checks.
Flip that around. Roughly 44% of the web did not pass.
That number even slipped a little from the month before, which is the first real dip after years of slow improvement. Not a crisis. But it tells you the easy wins are gone and a lot of sites are stuck.
The three things Google grades are simple to describe:
Loading (LCP) main content visible in under 2.5 seconds
Interaction (INP) taps and clicks respond in under 200 milliseconds
Stability (CLS) layout doesn't jump around as the page loads
You have to clear all three at once. Nail two and miss one, and the whole page reads as failing. That's why a site that feels fast to you can still come back red.
It's graded on your customers' phones, not yours
This is the part that trips everyone up.
That free speed test you ran is a lab test. It loads your site once, on a simulated connection, in a controlled room. Useful for debugging. It is not the score Google uses.
The real score comes from field data: actual visits from actual people on actual phones, collected over a rolling 28-day window. And it isn't the average visit. Google grades you at the 75th percentile, which means at least 75% of your visitors have to get a fast experience. The slowest quarter of your traffic, usually people on older phones and weaker signal, decides your grade.
There's a kicker on the interaction metric. It literally cannot be measured in a lab, because it needs a real human tapping and clicking through your page. So you can score a perfect 100 on a lab test and still fail in the field. That happens constantly.
One more thing worth knowing: when you fix something, it takes about four weeks to fully show up in the field data, because the window is rolling. There's no instant gratification here. Ship the fix and wait it out.
The dashboard that said "you're fine" is gone
If you half-remember a tidy little summary in your Search Console that gave you a thumbs up on page experience, you're not imagining it. Google retired that at-a-glance summary panel back in late 2024 to cut clutter.
The data didn't disappear. The detailed report still lives in there. But the simple verdict, the one a busy owner would actually glance at, is gone, and the deeper report is one most small business owners never open. So the change went unnoticed by almost everyone it affected. The signal that decides part of your ranking quietly moved somewhere you don't look.
Why your site is probably slow on mobile
Mobile trails desktop by about eight points across the whole web, roughly 48% passing on phones versus 56% on desktop. And here's the uncomfortable truth for anyone on a rented template builder: the platform you're paying for largely sets your ceiling.
Drag-and-drop builders load a pile of scripts, tracking, chat widgets, and popups by default. Every one of those hurts the interaction score, because that metric counts every tap, not just the first one. The more a template stacks on, the slower it gets to respond.
And you usually can't remove any of it. Hosted builders let you add code, but not strip out the platform's own bloat. So for a lot of small businesses, part of the score is outside your control by design. You can't fix what you can't edit.
The other big lever is where your site is hosted. The same site on a cheap shared host versus a properly configured one can see its server response time swing from over a second down to a couple hundred milliseconds. Often the fix isn't a redesign at all. It's the plumbing underneath.
The good news you didn't expect
At 55.9% passing, clearing all three checks puts you ahead of nearly half the web. Page experience is an official tiebreaker. When your content and a competitor's are otherwise even, the faster, steadier site wins the slot. On a competitive search, a fraction of a second can be the difference between page one and page two.
So this isn't a "you're doomed" story. It's a winnable margin most of your competitors are ignoring because they're staring at a green lab score that doesn't count.
Faster sites also just convert better. People don't wait around. Every study on this for the last decade points the same direction: when the site responds quickly, more visitors stick around and buy. The speed grade and the revenue move together.
Where this leaves you
If you take one thing from this: the score that matters is measured on your real visitors' phones, not on a test you run once. A green lab result and a failing field grade live side by side all the time, and the field grade is the one Google ranks on.
Figuring out which of the three checks is dragging you down, and whether it's your template, your widgets, or your hosting, is the actual work. That's diagnosable. Most sites are failing on one specific thing, not all of it.
This is the kind of thing we handle for clients so they never have to think about it. We build sites that pass on real phones because they're built right underneath, not patched on top. If you want to understand where your own site stands, subscribe for free at kief.studio. Members get our guides and resources, and the first conversation about your site is always free, no commitment.