CrUX Soft Navigation Metrics: Looking at Each SPA Page Navigation Separately

The Page Doesn't Reload, But the Reader Has Already Moved to a New Page

Single Page Applications can replace content and update the address bar while continuing to use the same document. Traditional field metrics might not necessarily attribute these subsequent pages as a new HTML navigation would. In her October 13 article, Annie Sullivan introduced experimental CrUX metrics that include both hard and soft navigations; it's built on the Soft Navigations API introduced in Chrome 151.

First, ask yourself if your site truly has a significant number of soft navigations. Figure 1 in the original article compares all CrUX origins with the top 1,000 sites: sources heavily reliant on soft navigations are a minority, and they are more common among these large sites. This is an aggregated observation from the publisher, not a measurement result from Boloop, nor a prediction for your site.

For SPAs with many subsequent page changes, including these pages will alter the measurement boundaries. LCP may include faster transitions that reuse already downloaded resources; INP and CLS are segmented by page fragment, rather than attributing all interactions or shifts from the entire application session to the initial landing page. Therefore, an improvement in experimental scores could come from changes in measurement methodology; you cannot rely on it alone as proof that a particular code optimization was effective.

How to Start Comparing

Compare the stable series and the experimental series for the same URL or origin, within the same date window. The experimental fields listed in the original article are experimental_largest_contentful_paint_with_softnavs, experimental_interaction_to_next_paint_with_softnavs, and experimental_cumulative_layout_shift_with_softnavs, available from the CrUX API, History API, and BigQuery dataset. First, verify the usual data eligibility and coverage; the absence of results for a specific URL is not necessarily a fault.

If you want to set up your own Real User Measurement, the related Chrome documentation explains how to attribute entries to the correct navigation and reset metrics between views. This is subsequent engineering work; this site's recommendation of this article does not mean we have already integrated these experimental metrics.

During the experimental evaluation, the stable Core Web Vitals scores in Google remain unchanged. The target of formalization mentioned in the article for May 2027 is still dependent on feedback and cannot be stated as if Search Console has already switched. It is suitable reading for those maintaining SPAs, or who need to explain why two performance reports differ; it is not a promise of search ranking improvements. The original article and charts belong to Annie Sullivan / web.dev; this site's text is an attributed reading guide.