Only 55.9% of origins pass Core Web Vitals in the latest field data, so "fast enough in the lab" is now a liability, not proof.
A vendor sends a green Lighthouse screenshot and calls the site fast. Real users load it on a mid-range Android phone on a weak connection and get a different experience. The gap between those two numbers is where directors lose credibility with leadership, because leadership eventually checks the field data.
The Green Screenshot Is Lying to You
The incumbent way reports lab scores. Lighthouse runs a simulated load in a controlled environment, produces a number, and the number goes into the deck. It is repeatable, it is easy, and it is not what your customers feel.
Field data measures real Chrome users over a rolling window. When the lab says green and the field says fail, the field is the one the business should believe, because the field is the business.
55.9% Pass, and That Is the Good News
Google's own CrUX release notes report that in the May 2026 dataset, published June 9, 2026, only 55.9% of origins had good Core Web Vitals, down 0.8%, across 18,445,974 origins (Chrome CrUX release notes). Google notes a regression in pass rates this month, mostly on Android, and says it has nothing definitive to share yet on the cause (Chrome CrUX release notes).
Almost half of the measured web fails the standard Google set, and the number moved the wrong way. The blogs claiming Google changed how INP is scored are wrong; the CrUX notes state no INP method change for May 2026, and describe a mostly-Android regression instead (Chrome CrUX release notes). Do not repeat that story to leadership.
One Weak Metric Sinks the Whole Page
Here is the part most speed programs miss. In the same dataset, 68.6% of origins had good LCP, 81.3% had good CLS, and 86.6% had good INP, yet only 55.9% passed all three (Chrome CrUX release notes). These are aggregate origin figures, not page-level, so segment before acting (Chrome CrUX release notes).
An 86.6% INP rate next to a 55.9% all-three rate tells the story. Most sites do not fail everywhere. They fail on one binding metric while the composite average looks reassuring. A program that optimizes the average chases the wrong target and leaves the failing metric in place.
Your SPA Was Never Fully Measured
Single-page apps have a deeper problem. Chrome's soft-navigations documentation confirms Core Web Vitals can now be measured per page navigation for single-page apps, that the feature is aimed to be on by default in Chrome 151, and that how CrUX will report soft navigations is still to be determined (Chrome soft navigations).
Read the implication. A SPA-heavy B2B site that "passed" may have been graded on the first page load only. Every in-app navigation after that went unmeasured. When per-navigation measurement lands, the honest baseline resets, and the site that looked fast gets a truer, harder grade.
The Fork You Now Face
You have two options for how you report speed. Keep sending lab screenshots and defend a number your customers do not experience. Or adopt field data as the source of truth, segment by device, and fix the single failing metric per template.
That is not a website that is fast. That is a screenshot that is green.
A lab score with no field data is decoration. A composite average with no per-metric view hides the binding constraint. A SPA grade with no soft-navigation measurement is a first-load illusion.
Report field data or stop reporting speed. Pull CrUX into leadership reporting, split the numbers by device so the Android regression is visible, and target the one metric that fails each template instead of grinding the composite. The teams that make this switch stop getting surprised. The teams that keep sending green screenshots get audited.
Speed is what your slowest real user feels, not what your fastest lab run shows.
Work With Magnet
Magnet builds digital experiences and conversion architecture measured against field data, per device and per metric, so the number in the report matches what customers feel. See how Magnet builds web that performs at https://www.magnet.co.
Sources
- Chrome CrUX release notes (May 2026 dataset): https://developer.chrome.com/docs/crux/release-notes
- Chrome soft navigations documentation: https://developer.chrome.com/docs/web-platform/soft-navigations
- Magnet: https://www.magnet.co
