CrUX for SPAs: How Chrome Measures LCP, INP, and CLS Across Soft NavigationsArticle · digest for October 30
European ligne claire BD comic illustrating a frontend developer watching CrUX metrics shift from red penalties to the green zone with SPA soft navigations.
Article4 sections≈ 7 min

CrUX for SPAs: How Chrome Measures LCP, INP, and CLS Across Soft Navigations

If you've spent the last few years building complex interfaces in React, Next.js, Vue, or Svelte, you've probably felt your heart sink at least once while staring at Core Web Vitals in Google Search Console. You could spend hours tuning performance, trimming bundles, and dialing in caching, yet Search Console stubbornly slapped yellow and red badges on your Single-Page Applications (SPAs) for sluggish interaction latency (INP) or layout shifts (CLS).

The worst part was that in real life, the app felt snappy: users clicked through catalog tabs, opened product cards instantly via the client router without a full page reload, and had a great experience. Yet the Chrome User Experience Report (CrUX)—the dataset Google feeds into search ranking signals—insisted on judging your app by Web 1.0 rules.

Now, that long-running mismatch has finally hit a turning point. Google Chrome's web platform team, led by Annie Sullivan, has officially rolled out experimental CrUX metrics with full support for Soft Navigations. Let's look under the hood: why SPAs were penalized for years, how the browser learned to spot route changes without a reload, and what this means for frontend teams.

The Architectural Bias: Why Classic CrUX Punished SPAs

To see how big this issue was, it helps to remember how Core Web Vitals were originally wired up. The entire CrUX telemetry model was built around Hard Navigations—the classic Multi-Page Application (MPA) flow where a user clicks a link, the browser sends a fresh HTTP request, parses a brand-new HTML document, builds the DOM from scratch, and paints the screen.

Every hard load counted as a separate, isolated Page View, and for each one the browser recorded three key metrics:

  • Largest Contentful Paint (LCP) — how fast the largest visible content block appears;
  • Interaction to Next Paint (INP) — how quickly the UI responds to clicks and keystrokes;
  • Cumulative Layout Shift (CLS) — how much the layout jumps around while loading.

In a Single-Page Application, life works completely differently. After the initial shell loads, every subsequent transition happens on the client: the router intercepts the click, updates the URL via the History API (pushState or replaceState), fetches a small JSON payload, and swaps new components straight into the existing DOM. The browser never reloads the tab.

That's where the old CrUX made a fatal logical mistake: it treated an entire 20-minute user session as a single Page View! The distortions were brutal:

  1. Interaction latencies (INP) accumulated across the whole session. If a user browsed 50 products and a heavy dropdown on item #49 took 250 ms to respond, that worst-case click got pinned onto the initial landing URL.
  2. Layout shifts (CLS) snowballed. Every trip to the cart or filter drawer tossed tiny shifts into the same bucket, turning a green score hopelessly red.
  3. Worst of all, LCP for subsequent screens was ignored altogether. The speed of opening the second, third, or tenth route inside an SPA didn't exist in Core Web Vitals—only the first, heaviest cold load with all scripts and stylesheets was measured.

Because of this, plenty of teams abandoned client-side rendering and rushed to rewrite their apps for Server-Side Rendering (SSR) or plain MPAs just to keep Google Search Console happy.

The Soft Navigations Heuristic: How Chrome Detects Client-Side Transitions

Fixing this required a formal web platform standard. In Chrome 151, the Soft Navigations Web Platform API stabilized, teaching the browser engine to detect screen transitions automatically.

But how does a browser tell a real route change apart from opening a modal, switching a tab, or toggling dark mode? The W3C working group defined a strict chain of heuristics. A transition counts as a soft navigation only when all three conditions hold:

  1. It is initiated by a direct user action: a mouse click, touch tap, or key press.
  2. During that interaction's handling, the URL changes via history.pushState(), history.replaceState(), or the native Navigation API.
  3. Significant DOM modifications occur: an old content node is removed and a new content block with real text or graphics is mounted in its place.

Once that chain completes, the browser marks a new navigation boundary: CLS counters reset for the new screen, a fresh LCP window opens, and INP starts attributing interactions to the new route.

In our walkthrough of how a practical guide to the Performance panel in Chrome DevTools for debugging Core Web Vitals works, we already saw these boundaries appear on the Interaction & Timings track. Now that exact same logic powers global telemetry across millions of users.

The CrUX Experiment: Numbers That Change Everything

Starting in October 2026, the Chrome User Experience Report (CrUX) gained three new experimental metrics:

  • experimental_largest_contentful_paint_with_softnavs
  • experimental_interaction_to_next_paint_with_softnavs
  • experimental_cumulative_layout_shift_with_softnavs

According to the announcement on Chrome for Developers, the first real-world traffic numbers landed like a bombshell.

For sites with a high share of soft navigations (where more than 75% of transitions happen on the client), the percentage of origins achieving a "Good" LCP on Android jumped from 26% to a stunning 87%! Let that sink in: nearly three-quarters of mobile SPAs were flagged as "slow" for years simply because the measurement tool ignored how fast they ran after the first load. On subsequent SPA transitions, the browser doesn't have to re-download and re-parse heavy shared CSS, fonts, and framework bundles—they're already warm in memory, so the next screen renders in a fraction of a second.

Responsiveness metrics saw a similarly dramatic shift. Because long sessions no longer pile every click penalty onto one URL, the share of sites passing the green INP threshold nearly doubled. And per-URL breakdowns in the CrUX API finally let developers pinpoint bottlenecks: instead of Search Console blaming the home page, the report clearly shows that latency spikes come from a heavy cart component at /checkout.

How to Access the Data and the Roadmap to May 2027

How can you inspect these metrics for your own app right now? You already have four ways in:

  1. CrUX API: the experimental fields are returned in standard API responses for any domain or URL.
  2. Google BigQuery public dataset: the chrome-ux-report.experimental.soft_navigations table lets you run cohort queries and benchmark against competitors.
  3. CrUX Vis: the visualization tool at cruxvis.withgoogle.com now includes a one-click toggle comparing the official score with the soft-navigation numbers.
  4. web-vitals library: version 4.x supports soft-navigation attribution so you can pipe these events into your own analytics (GA4, Sentry, Datadog).

In our look at how new Chrome 156 and web standard features are reshaping the frontend platform, we noted how browser primitives are catching up with modern frameworks. Soft Navigations is one of the clearest examples yet.

Still, keep a level head: these metrics are currently experimental. By the Chrome team's official schedule, the experimental window runs until May 2027. Until then, Google Search Console and search ranking algorithms will continue using the old formula without soft navigations.

Even so, you and your team now have an 18-month runway to audit your web apps, verify your History API usage, and head into May 2027 with clean green scores.

Want a deep diveVotes help us choose what to break down in detail. This material has 0 votes.
Sources · 2