CrUX adds experimental soft-navigation metrics without changing stable Core Web Vitals

Google made experimental soft-navigation measurements available in the Chrome UX Report, or CrUX, on October 13, 2026. The release announcement says the new data accounts for page-like transitions that change content and the URL without a full browser navigation. Stable Core Web Vitals measurements remain unchanged.

That makes this a measurement development, not evidence that a site's search performance suddenly improved or deteriorated. For an owner of a JavaScript-heavy publication or web application, the immediate opportunity is to examine experiences that a single initial page load does not fully describe.

One browser load can contain several reader journeys

A conventional navigation loads a new document. In a single-page application, the next article or view can instead be rendered inside the existing document. Google's soft-navigation documentation explains the browser's approach to identifying these transitions; not every visual update should be treated as a new page.

The new experimental measurements include Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS) with soft navigations. Google's announcement describes measurement across the separate views rather than attributing the whole sequence only to the initial navigation.

For a publisher, a useful test case is a reader opening an article, following an internal link and then opening another view without a full reload. The question is whether the later view feels slow or unstable, even when the initial entry page looked acceptable.

Keep the experimental and stable series apart

The release makes the experimental data available through CrUX's data products, including its APIs, BigQuery dataset and CrUX Vis. The CrUX API documentation is the place to check the requested metric names and the response your monitoring actually receives.

Do not overwrite a historical stable series with an experimental series and call the difference a performance regression. Label the measurement method, compare like with like and retain the original series while you assess the additional data. A changed definition can change a number without a new site release.

Google's Search Console Core Web Vitals documentation remains the reference for the production report. The October announcement does not switch that report to these experimental measurements. Google describes May 2027 as an intended graduation point, subject to feedback, not a guaranteed deadline or a promise of ranking gains.

Use the data to choose a reproducible investigation

Start with a route readers actually follow, then reproduce its transition on a supported browser and device. Record what changes, whether the URL changes, what becomes the main content and which interactions remain responsive. The investigation should end with a specific view or behavior, not a site-wide accusation based on one unfamiliar metric.

Where traffic has also changed, keep performance diagnosis separate from acquisition diagnosis. Our traffic-drop investigation guide explains why an observed decline needs its own evidence; these experimental measurements cannot identify its cause on their own.

Add a small set of representative navigation checks to the site's maintenance routine. The useful outcome is a clearer picture of what happens after the first page, while stable reporting remains available for continuity.