Core Web Vitals in 2026: Why INP Is the Metric That Matters

In March 2024, Google replaced one of its main site-speed measurements (FID) with a tougher one called INP. Two years later, the impact is everywhere: sites that used to pass Google's speed checks are now failing them, often without changing a single line of code. The difference comes down to what the new metric actually measures.

The old metric only checked how quickly your site responded to a visitor's first click. The new one checks every click, tap, and keystroke during the whole visit, and reports the worst one. So sites that load fast but feel sluggish after they load, common on modern JavaScript-heavy sites, are exactly the sites INP catches out.

The Current Thresholds

As of 2026, the three Core Web Vitals are:

• LCP: how fast the main content shows up. Good means under 2.5 seconds for most visitors.
• INP: how responsive your site feels when people interact with it. Good means under 200 milliseconds for most visitors.
• CLS: how much your layout jumps around as the page loads. Good means very little visual jumpiness.

That 200-millisecond bar for INP is much harder to hit than the old metric's bar, because it judges every interaction during the visit instead of just the first one.

What Actually Causes Slow Sites

In our experience, and confirmed by published industry data, the four most common causes of INP failures are:

• Heavy third-party scripts, chat widgets, marketing tags, A/B testing tools, running at the wrong moment and blocking your site from responding.
• Modern app frameworks re-rendering huge sections of the page in response to small interactions, like a search box updating the entire layout on every keystroke.
• Analytics and tag manager scripts that fire too aggressively. A Google Tag Manager container with dozens of tags can add hundreds of milliseconds to the first interaction.
• Modern server-rendered sites still have to 'wake up' in the browser. On large pages, this wake-up cost shows up as poor INP for visitors who interact early.

How to Find the Problem

Use Chrome's built-in Performance tools and Google's PageSpeed Insights. The most important thing: trust real visitor data (what Google calls CrUX) over the lab-test scores from Lighthouse. Lab tests miss real-world responsiveness problems because they don't simulate the way real people use a site.

Google Search Console groups slow pages together by URL pattern, which makes it easy to see which areas of your site are dragging down performance. Usually, fixing the worst-performing template lifts the whole site.

What Actually Fixes It

• Delay non-essential third-party scripts until people actually need them. Most chat widgets and marketing tools have a setting to load only after the visitor scrolls or interacts.
• Audit your Google Tag Manager setup. Many tags don't need to fire instantly, move them to delayed triggers or remove ones nobody actually uses.
• If you're on a modern React or Vue site, use the framework's built-in tools (React's startTransition, for example) to mark non-critical updates as deferrable so urgent interactions stay snappy.
• Break up long-running JavaScript work into smaller chunks so the browser can stay responsive between them.
• For content-heavy sites, look at modern frameworks (Astro, React Server Components) that only 'wake up' the parts of the page that actually need to be interactive.

Why This Matters Beyond Google Rankings

Google rankings matter, but they're not the main reason to fix this. INP correlates directly with how 'broken' your site feels to visitors. A site that lags every time someone clicks feels unprofessional regardless of where it ranks. The same work that earns a good INP score also makes the site feel sharper and converts better.

Want to Hear More?

Conversations about how technical craft, performance, and digital operations actually show up in mission-driven work happen regularly on The Nonprofit Syndicate Podcast, which we're a part of. You can find it on YouTube at The Nonprofit Syndicate Podcast channel.

Sources

• Google web.dev, Interaction to Next Paint (INP) official metric documentation. web.dev/articles/inp
• IncRev, Core Web Vitals Study 2025: Analysis of 208,085 Websites. increv.co
• Mlopez.ca, Core Web Vitals 2025: INP Optimization Guide, February 2025. mlopez.ca