Skip to content

Contact

Mexico City, Mexico
Available for remote and on-site projects.

Professional networks

LinkedIn · GitHub · Email
Applied engineering, measurable results.

November Core Update and React sites

· 6 min read · Samuel Torres

By Samuel Torres — Software & Cloud Engineer. 23+ production sites.

Two things landed this month that matter to anyone running a React or Next.js site. On November 10, Google started rolling out the November 2026 Core Update. On November 11, the October CrUX release came out (the Chrome User Experience Report, where Search Console's Core Web Vitals report gets its data), and its APIs now carry Core Web Vitals with soft navigations: each view change in a SPA is measured as a page of its own. One moves rankings, the other changes how your pages are measured; keep them apart. Here's what I'd check if your site runs React.

What is actually confirmed about the Core Update

On the Google Search Status Dashboard, the "November 2026 core update" incident reads, dated November 10, 2026: "Released the November 2026 core update, which applies globally and to all languages. The rollout may take up to 2 weeks to complete." As of November 20 the dashboard does not mark it as finished: it's still rolling out, with no official end date.

Google's documentation on core updates says to confirm the rollout has finished, wait at least a full week before analyzing your site in Search Console, and compare that week with one before the update started. If your traffic moved on the 12th, you don't know anything yet.

Google doesn't say this update has anything to do with Core Web Vitals, and I won't draw that line for it. But while checking your site you'll open the Core Web Vitals report, which is fed by CrUX, and that's where a React site needs to understand what changed.

What changed in CrUX: soft navigations

A SPA (React Router, Next.js's <Link>) doesn't load a fresh HTML document on every click: it swaps content with JavaScript and updates the URL. That's a soft navigation. Until now CrUX only saw the initial load (the "hard navigation") and charged every interaction and layout shift in the session to the landing page.

Chrome defines a soft navigation with three conditions: it is started by a user action, it changes the visible URL, and it produces a visible paint. The Soft Navigations API is enabled by default from Chrome 151.

In the September 2026 CrUX release (published October 13), Google added, in its own words, "new experimental variants of the Core Web Vitals (LCP, INP, and CLS) including soft navigation metrics as used by SPA sites," available in BigQuery and the CrUX APIs. They are called:

  • experimental_largest_contentful_paint_with_softnavs
  • experimental_interaction_to_next_paint_with_softnavs
  • experimental_cumulative_layout_shift_with_softnavs

The October 2026 release (published November 11) added FCP and navigation types with soft navs to the APIs, and the global number looked like this: with the stable metrics, 56.3% of origins had good Core Web Vitals; with soft navigations included, "60.3% of origins had good Core Web Vitals (an ↑ 7.1% from when soft navs are not included)." Read that carefully: 7.1% is a relative increase (56.3 to 60.3 is 4 points), not 7.1 points.

From the web.dev announcement, two details tell you whether this affects you:

  • Around 70% of origins have little to no soft navigation traffic (under 5% of page views), and only 7–11% use soft navigations for more than half of their page views. Among the top 1,000 sites, though, it's 35–42%.
  • Google aims to move these metrics out of the experimental stage in May 2027, subject to feedback, and says: "Once graduated, Search Console will switch to use the new data."

Search Console still uses the stable metrics; the experimental ones show you, months ahead, how your site will look.

What it means for a React or Next SPA

According to the announcement, on sites with lots of soft navigations:

  • LCP improves, because follow-up views reuse code and assets already downloaded (SPAs aren't faster overall).
  • INP improves, because it's measured per view, not as the slowest interaction of the session. But Google warns these sites still trail multi-page sites on "good" INP, and heavy JavaScript during route transitions still causes problems.
  • CLS improves, because layout shifts are split per view instead of piling up across the session.

Also, the CrUX API reports each soft navigation under the destination URL, not the landing page: a slow /product/[id] can now show up under its own URL, if it has enough data.

My read: if your numbers improve with soft navs, you didn't fix anything; it's a fairer measurement. If INP is still "needs improvement" or "poor," the problem is real and lives in your route JavaScript.

What to check this week

1. CrUX Vis

Open CrUX Vis, enter your origin (or a specific URL) and compare each stable metric against its soft-nav variant, phone and desktop separately. You're looking for:

  • A big gap between stable and experimental? Your site has lots of soft navigations, and your report will change when Search Console adopts the new data.
  • INP with soft navs still above 200 ms? The new measurement won't fix that.

2. The CrUX API, if you want exact numbers

With a Google Cloud API key you can request both variants in one query:

curl -s --request POST \
  'https://chromeuxreport.googleapis.com/v1/records:queryRecord?key=API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{"origin":"https://tutienda.mx","formFactor":"PHONE",
    "metrics":["interaction_to_next_paint",
               "experimental_interaction_to_next_paint_with_softnavs"]}'

Swap origin for url to query a single page. Each metric in the response comes with its histogram and its 75th percentile (percentiles.p75), which is the value used for the rating.

3. Search Console's Core Web Vitals report

This report's data comes from CrUX: real users, not a test run. It groups URLs with a similar experience and rates them at the 75th percentile: LCP good up to 2.5 s, INP good up to 200 ms, CLS good up to 0.1. If a URL group doesn't have enough data, it isn't shown on its own. And when you fix something and start validation, Search Console opens a 28-day monitoring session, so don't expect green the next morning.

Why a green Lighthouse score isn't enough

Lighthouse is a lab test: one load, simulated device and network, nobody clicking. No real interactions means no real INP; web.dev's INP guide says Total Blocking Time can be a reasonable proxy, not a substitute. And in a SPA the trouble shows up after load: opening a menu, filtering, changing routes.

Simon Janvier puts it plainly in his practical Core Web Vitals 2026 guide: "A site can show an excellent local Lighthouse score while still scoring "poor" on INP in production." Decide with field data and keep Lighthouse for diagnosis. For WordPress I covered it in optimizing Core Web Vitals in WordPress.

How to measure real INP on your React site

CrUX won't tell you which route or interaction is failing; for that, measure on your own site. Google's web-vitals library (version 6 and up) reports metrics per soft navigation with the reportSoftNavs: true option. In Next.js with the App Router, a client component mounted once in the root layout:

// app/web-vitals.jsx
'use client';
import { useEffect } from 'react';
import { onCLS, onINP, onLCP } from 'web-vitals';

function send(m) {
  navigator.sendBeacon('/api/vitals', JSON.stringify({
    name: m.name,
    value: m.value,
    rating: m.rating,
    id: m.id,
    navigationType: m.navigationType, // 'soft-navigation' for SPA views
    url: m.navigationURL,             // the URL this metric belongs to
  }));
}

export function WebVitals() {
  useEffect(() => {
    onINP(send, { reportSoftNavs: true });
    onLCP(send, { reportSoftNavs: true });
    onCLS(send, { reportSoftNavs: true });
  }, []);
  return null;
}

/api/vitals is your own endpoint (a Next route that stores the JSON in your database or logs). Three things matter:

  • Use navigationURL, not location.href: a metric can arrive after the user has already moved to another route.
  • Segment by navigationType. soft-navigation views behave differently from the initial load.
  • In browsers without the API (non-Chromium, or Chrome before 151), reportSoftNavs changes nothing. If you want to keep comparing against your history, also register the callbacks without the option, as Chrome's documentation suggests.

Back up and test in staging first: every visit sends several metrics to your endpoint.

The real case

I'll be straight with you: I don't have a documented React SPA case with before-and-after CrUX data including soft navigations. The metric came out weeks ago, and I'm not going to make numbers up.

The case I do have is Conqr, and it isn't React: I migrated it from static HTML to WordPress. The home page went from 3.99 to 766 requests per second, with CLS 0, in 9 days. I bring it up for the approach, not the stack: the result was reported with measured numbers (requests per second on the home page and CLS), not a Lighthouse score. That's the same standard I'd hold a SPA to: decide with measurements of what users get, not with a green lab run.

If you're moving a WordPress site to React or Next, the SEO side is in how to migrate WordPress to React without losing SEO. What's new is that the site will be measured per view.

What I wouldn't do

  • Draw conclusions about the Core Update before Google marks it finished and another week goes by.
  • Blame Core Web Vitals for a traffic drop just because it lined up with the update. Google doesn't connect the two.
  • Rewrite a SPA as a multi-page site because of this. Measure INP per route first and fix the one that's bad.
  • Treat Lighthouse as the final proof that "we pass."

Want to go over it together?

If your site runs React or Next and you want to see how it looks in CrUX with and without soft navigations, and which routes have the worst INP, let's review it together. Reach out through contact with the URL.

Sources

About the author

Samuel Torres is a Software & Cloud Engineer at Conqr Digital Agency (Mexico City). He runs 23+ production WordPress sites and processed 80,000 documents with OCR + LLM. More at /en/about. Get a quote on WhatsApp in 2 min

Samuel Torres Pichardo

Samuel Torres Pichardo
Software and cloud engineer in Mexico City, with about 3 years of professional experience.
About me · LinkedIn · GitHub · Contact

Got a project in mind?

Hire me: I'll help you build it.

Websites, online stores, custom systems and cloud. Tell me what you need and I'll tell you how I'd build it, what it costs and how long it takes.

  • Mexico City · remote

Menu

Español

[email protected]