INP is coming: what changes with the new Core Web Vital

Google has confirmed that Interaction to Next Paint (INP) will replace First Input Delay (FID) as a Core Web Vital in March 2024. Since Core Web Vitals influence Google Search rankings, website owners should understand what this shift means and how to prepare before the change takes effect.

The current Core Web Vitals measure three aspects of user experience: loading speed via Largest Contentful Paint (LCP), interactivity via FID, and visual stability via Cumulative Layout Shift (CLS). This March, INP takes over as the interactivity metric. It is a stricter and more thorough measure of responsiveness, and the data from Cloudflare's network shows that many sites scoring well on FID will not score as favorably on INP.

Understanding the metrics: FID vs. INP

Both metrics measure delays around user interactions, but they capture different portions of the interaction timeline. An interaction can be broken down into three serialized phases: input delay (time until the browser can begin processing event handlers), processing time (time for the event handlers to run), and presentation delay (time until the next paint occurs).

FID measures only the input delay portion of the page's first interaction. Subsequent interactions are ignored. A page scores 'Good' FID if the input delay is under 100 ms, 'Needs Improvement' between 100 and 300 ms, and 'Poor' above 300 ms.

INP measures the full duration—input delay, processing time, and presentation delay—for every click, tap, and keyboard interaction during a page view. The reported value is the longest interaction observed, excluding outliers. The thresholds are stricter in practice: under 200 ms is 'Good', 200 to 500 ms is 'Needs Improvement', and anything above 500 ms is 'Poor'.

This distinction matters. A single interaction on a page shows the difference: FID records just the input delay, while INP captures the complete time until the next paint. With INP, sites must ensure that all interactions throughout a visitor's session stay within the 'Good' threshold, not just the first one.

What the data from 850,000 sites shows

Cloudflare's Real User Measurements (RUM) infrastructure already collects INP data, producing a dataset spanning hundreds of thousands of sites. Analyzing a week of traffic across these sites—over four billion reported FID and INP values—reveals a significant gap between the two metrics.

93% of FID values classified as 'Good', while only 75% of INP values did. The distribution difference is even more striking at the percentiles: the P75 value for FID is just 16 ms (0.16 times the FID 'Good' threshold), while the P75 for INP sits at 200 ms—exactly at the INP 'Good' cutoff.

Device type also plays a role. When breaking INP data down by device, 88% of desktop INP scores classified as 'Good', but only 67% of mobile INP scores did. Mobile interactivity is clearly where sites are struggling most.

Preparing your site for the switch

Because INP captures the complete interaction lifecycle across an entire visit, improving it means looking beyond initial page load. The challenge is that INP is best measured with field data—observations from real users on real browsers. Synthetic testing tools like Google Lighthouse are not well suited for this, as they cannot emulate the broad range of browsers, operating systems, and user behavior patterns that field data captures.

To collect real user data, sites need a Real User Monitoring (RUM) provider. This is the necessary first step toward understanding how a site's INP actually performs. Once data is flowing, analyzing it reveals which pages have interactivity issues. Optimization can then target the three components of INP: reducing input delay, shortening processing time, and speeding up presentation delay.

Optimization tactics include breaking up long JavaScript tasks, minimizing DOM manipulations and layout thrashing, avoiding timers, and limiting third-party code. These measures are not trivial to implement, so starting early is the safest strategy.

Tooling and mitigations

Cloudflare offers a free RUM provider that collects INP alongside its other performance data. It powers Cloudflare Web Analytics and Cloudflare Observatory, and can be enabled from the Cloudflare Dashboard under Web Analytics.

One frequent cause of poor INP is heavy third-party JavaScript that attaches event listeners to interactive elements. These scripts block the main thread after user interactions, inflating processing time. Cloudflare Zaraz is designed to remove third-party scripts from the browser entirely. This frees up the main thread and can improve both INP and the site's security posture by reducing the amount of third-party code executed on the page.