From 39% to 55.3%: Four years of Core Web Vitals progress, and why the number never sits still

In our last piece on this topic, we cited data showing a 39% Core Web Vitals pass rate for the web overall.

Four years later, Google's newly published Chrome UX Report data for June 2026, released July 14, puts the figure at 55.3% of origins passing all three Core Web Vitals. That's a 16-point climb in four years. The web, broadly, is measurably faster than it was. The question that actually matters for your store isn't whether the internet improved. It's whether you did.

Four years, 16 points: What actually moved the needle

The rise from 39% to 55.3% wasn't a straight line, and it wasn't driven mainly by individual site owners chasing a score. Reading through four years of Google's own release notes, most of the biggest single-month jumps came from browser-level and platform-level changes that lifted huge swathes of the web at once, whether those sites were paying attention or not.

• In January 2022, Chrome's broad rollout of the back/forward cache alone lifted the global pass rate by 7.4% in a single month.
• In September 2023, a round of LCP image improvements shipped in WordPress 6.3, a platform that accounts for roughly a third of the entire CrUX dataset, and the effect showed up in the aggregate numbers within weeks.
• By August 2024, the industry crossed the symbolic halfway point, with over 50% of origins passing, helped along by an INP optimisation Google shipped for its own ad-serving library that lifted mobile INP pass rates for ad-heavy sites from roughly 56% to 66% almost overnight.

Even the definition of "passing" moved. When Google formally replaced First Input Delay with Interaction to Next Paint as the responsiveness metric in March 2024, the pass rate recalculated under the new, stricter metric came in several points lower than the old one, a reminder that the goalposts themselves aren't fixed.

As recently as April 2025, a single CMS vendor, Duda, removed a caching header from roughly 40,000 of its origins and nudged the entire dataset's CLS and LCP pass rates up by a percentage point or more. One vendor's config change moved a global statistic. That should tell you something about how little control any individual store actually has over the ground shifting beneath it.

The number moves every month, not just every four years

The four-year trend line is real, and it's worth feeling good about. But it flattens something important: month to month, the number is genuinely volatile, and often for reasons even Google can't fully explain.

The June 2026 release, the one that put the pass rate at 55.3%, was actually a 1.2-point drop from May's 55.9%, which itself was down 0.8 points from April's 56.4%. Google's own commentary across the past year reads like a running list of unexplained dips and seasonal shrugs: a mobile-specific regression in May 2026 with "a few suspicions but nothing definitive," a data quality issue affecting Android devices through mid-2025, and a near-annual note that December data swings unpredictably for reasons that have never been fully pinned down.

None of that is a criticism of Google's methodology. It's the opposite point: if the organisation that owns the metric can't always explain a month-to-month swing, no individual store should assume their own pass rate, checked once last quarter, still holds true today.

Are you part of the 55%?

That's the real question behind this update. Not "is the web getting faster" (it is, on aggregate), but "where does my store sit right now, this week, on the devices and networks my actual customers are using?"

The 55.3% figure is a global average across all origins, not a target you hit once and keep. Given how much the number moves release to release, much of it for reasons outside any one site's control, the only honest answer to "are you part of the 55%" is one you check continuously, not one you assume from a Lighthouse run three months ago.

The security-performance intersection nobody talks about

Here's one of the quieter reasons a passing score doesn't stay passing: security tooling overhead rarely makes it into Core Web Vitals discussions, but it directly affects them.

eCommerce sites face an escalating threat landscape. Sansec's recent discovery of a WebRTC-based payment skimmer, one that bypasses Content Security Policy to steal payment data, illustrates how sophisticated attacks have become. In response, stores layer on security scripts, integrity checkers, bot detection, and fraud prevention tools.

Every one of those tools adds JavaScript to your pages. Every byte of JavaScript competes for main thread time. Every millisecond of main thread time directly impacts your INP score. A store that passed Core Web Vitals in January can quietly fail by June simply by doing the responsible thing and adding a new fraud-prevention script, with nobody connecting the two events.

From a snapshot to a live number

If Google's own aggregate figure, drawn from millions of origins, swings by more than a point most months for reasons even its own team can't always name, a single quarterly audit on your own store is measuring a moment that may not survive the week.

This is the operating model AuditIQ eCommerce monitoring tool enables. Rather than treating Core Web Vitals as a box to check once, AuditIQ monitors them continuously, alongside the business-critical user journeys they actually affect, so a regression shows up as it happens, not in next quarter's audit.

When a new security script pushes your mobile LCP past the 2.5-second threshold, you see it in context: this script, on this page, affecting this percentage of real users, with this estimated revenue impact. When a platform update or a new app quietly changes your INP, you know within hours, not whenever someone happens to re-run Lighthouse.

The bottom line

The web went from 39% to 55.3% in four years, and that's a genuinely good story about the internet getting faster. But an improving industry average doesn't help a store that regressed last month and hasn't checked since. Passing today says nothing about passing next month, and the data backs that up release after release.

The stores that stay on the right side of that number long-term won't be the ones that passed one audit. They'll be the ones who stopped treating performance as a quarterly project and started treating it as something they watch, continuously, the same way the number itself keeps moving.

Start measuring what matters. Try AuditIQ for free and see how AuditIQ turns Core Web Vitals from a monthly surprise into a daily, continuously monitored number.