Google adds experimental soft navigation Core Web Vitals to CrUX

Chrome has started publishing experimental versions of LCP, INP and CLS, the three Core Web Vitals, that include "soft navigations": cases where a single-page application uses JavaScript to change what is on screen and update the URL without the browser loading a new page. The new metrics landed in the Chrome User Experience Report (CrUX) on 13 October, in the BigQuery dataset, the CrUX API and the CrUX History API, and CrUX Vis now lets you compare old and new numbers for your site.

Search Console stays on the stable numbers for now, and Chrome says that will only change once the new metrics graduate, which it is targeting for May 2027, subject to feedback. So for now this is something to look at, not something to act on.

What a soft navigation is

Most websites load a fresh HTML page from the server on every click. A single-page application (SPA) works differently. It loads once, then uses JavaScript to swap out the page content and update the URL in the address bar without a full reload. Chrome calls that transition a soft navigation. React, Vue or Angular front ends, headless CMS builds, web apps, booking tools and ecommerce sites often work this way.

Chrome's definition has three parts: the navigation is started by a user action, it results in a visible URL change, and it results in a visible paint. Detection is built into Chrome and on by default from version 151, which reached stable in late July 2026 with the Soft Navigations API. The new CrUX data builds on that launch.

The measurement gap this closes

Until now, CrUX only recorded the first "hard" page load on an SPA. Every subsequent in-app view change was ignored as a page load, and every interaction and layout shift from the whole session was attributed back to that first landing page. Chrome's blog post describes SPAs as a gap for the Core Web Vitals initiative. Each framework handles these transitions differently, so measuring them meant custom code that was hard to compare across sites.

Now every in-app view change that alters both the content and the URL counts as a separate navigation. The CrUX API records those results against the page the user moved to, not where the visit started, provided that page meets CrUX's normal eligibility rules. That makes it easier to see which screen in an app is slow.

The three new metrics

The new fields sit alongside the stable ones and measure the same three things:

• experimental_largest_contentful_paint_with_softnavs measures Largest Contentful Paint (LCP) across both hard and soft navigations.
• experimental_interaction_to_next_paint_with_softnavs measures Interaction to Next Paint (INP) per soft navigation as well as per hard navigation.
• experimental_cumulative_layout_shift_with_softnavs measures Cumulative Layout Shift (CLS) per individual view and URL.

Chrome expects to add experimental soft-navigation versions of First Contentful Paint and navigation types in the next CrUX release in November.

Most sites will see little or no change

Depending on platform, 66 to 75 percent of all CrUX origins see soft navigations on fewer than 5 percent of page views, while they make up the majority of views at just 7.0 percent of origins on Android and 10.6 percent on desktop. Chrome's advice is blunt: if your site does not use JavaScript to swap content and update the URL without a reload, you should not expect any change in your metrics.

The picture is different at the top of the web. For 35 percent of the 1,000 most popular sites on Android, and 42 percent on desktop, soft navigations make up most page views.

On most standard WordPress, Shopify or Squarespace sites, each click loads a new page, so expect little or no change. Some themes and headless builds do swap pages with JavaScript, and Wix is moving its sites from single-page to multi-page navigation, so some Wix sites may still see a difference. The quick click test below will tell you. For a custom React or Vue front end, a headless build, or a web app behind a login, the experimental numbers may differ a lot from the stable ones.

Where the scores change

The gains are concentrated in sites that use soft navigations for most of their page views, where Chrome's figures show all three metrics improving, some sharply.

LCP. On Android, origins that serve 75 percent or more of their views as soft navigations jump from 26 percent rated "good" on stable LCP to 87 percent on the experimental version. Chrome says this is not because SPAs are faster than multi-page sites overall. Follow-up clicks are faster on any site, because code and assets were already downloaded on the first load. The new metric simply brings SPAs with many follow-up views into line with multi-page sites with many follow-up views.

INP. Chrome now scores responsiveness for each page view rather than taking the slowest interaction from a whole visit. For sites built mostly on soft navigations, it says that roughly doubles the share that pass INP on Android. Those sites still lag multi-page sites on INP, though, and Chrome warns that heavy main-thread JavaScript when the app changes route may still cause INP problems.

CLS. Layout shifts are now counted per page view instead of piling up over a long visit. On desktop that puts SPA-heavy sites level with multi-page sites, and on Android they come close.

Across the whole dataset the effect is modest. The September 2026 CrUX release notes show 55.9 percent of origins passing all three Core Web Vitals on the stable metrics, and 59.5 percent with soft navigations included. Good LCP moves from 68.3 to 72.1 percent, good CLS from 81.7 to 85.1 percent, and good INP from 85.4 to 87.4 percent.

Why nothing changes in rankings yet

The stable scores remain unchanged while the experimental versions are evaluated alongside them, and until the metrics graduate the Search Console Core Web Vitals report stays on the stable numbers.

Chrome also warns that the measurement itself is still moving. Each metric is labelled experimental because the team is collecting feedback through the soft navigations specification repository, and it says to expect some month-to-month shifts as measurement improves. Chrome also notes that not every user will report soft-navigation metrics, since other browsers and versions of Chrome before 151 do not support the feature.

What is still unclear

The announcement does not say which stage follows experimental. Chrome's published Web Vitals lifecycle normally holds a metric at the "pending" stage for six months or more before it becomes stable, but Chrome has not confirmed these metrics will follow that route. Nor has it said whether the current "good" thresholds will stay once soft navigations are counted, or exactly how the Search Console switch will be handled. The May 2027 date is a target, not a commitment.

What to check this week

Start by working out whether your site is a single-page application at all. Click through a few pages. If the page visibly reloads on each click, you are on a conventional multi-page site and need only keep watching your stable Core Web Vitals in Search Console. If the URL changes but the page does not reload, or you are not sure, ask your developer whether the front end is React, Vue, Angular or a headless build.

If your site is an SPA, open CrUX Vis, enter your origin, and compare the experimental and stable versions of LCP, INP and CLS. The difference shows how your scores change once later views in a visit are measured on their own, rather than ignored (LCP) or piled onto the landing page (INP and CLS).

Pay particular attention to INP. It is the metric where SPA-heavy sites still trail multi-page sites under the new measurement, and Chrome points to heavy JavaScript during route changes as a likely cause. If your experimental INP is still outside the good range, that is a real problem the new metric will not solve.

Crawler research from December 2024 found that most AI crawlers, AI search bots included, did not render JavaScript (Google's Gemini, built on Googlebot's infrastructure, was the main exception), so content shown only after client-side rendering or a soft navigation may be invisible to them. Keep critical content, such as product details, titles and navigation, in the server-rendered HTML.

If you use a real user monitoring (RUM) tool, ask the vendor whether it supports measuring Core Web Vitals per soft navigation. Chrome's documentation says many providers are planning to test the new standard, and some already offer limited measurement based on their own heuristics.

Finally, do not rebuild a working multi-page site as a single-page app to chase these numbers; Chrome has said the LCP gain is not because SPAs are faster. Make a diary note for the November CrUX release, when FCP and navigation types are expected to join the experimental set, and for May 2027, when Chrome aims to end the experiment.