INP is the metric that actually matters now
Interaction to Next Paint has been the Core Web Vitals metric quietly tanking small business sites since 2024. Here's what it is, how to measure it, and the three things that fix it.
If I audit a small business site in 2026 and one Core Web Vitals metric is red, it’s almost always INP. Not LCP. Not CLS. INP.
Most owners have never heard of it. Worth five minutes to explain what it is, why it matters, and the short list of things that fix it.
What INP measures
Interaction to Next Paint is, roughly, how long it takes between you tapping something and the page visibly reacting. A click, a tap, a keypress — the browser measures the delay before anything on the screen changes in response.
Google’s thresholds:
- Under 200ms is good. Feels instant.
- 200–500ms needs work. You’ll notice a hitch.
- Over 500ms is bad. The page feels broken.
It replaced First Input Delay in 2024 because FID was too easy to pass. FID only measured the very first interaction; INP measures all of them.
Why your site is probably failing it
A site with a slow INP is almost always a site running too much JavaScript on the main thread.
The usual suspects, in rough order of how often I see them:
- Three or four analytics scripts, each adding their own event listeners.
- A chat widget (Intercom, Drift, Crisp) loaded on every page instead of only where it’s needed.
- A cookie banner that blocks interaction until it finishes initialising.
- A heavy framework hydrating components that don’t actually need to be interactive.
- A carousel or animation library that runs on every scroll event.
Each one is justifiable in isolation. Together, they mean every tap waits for the main thread to clear. On a mid-range Android phone — which is what most of your users are on — that can easily cost half a second.
How to measure it honestly
Lighthouse is not the right tool. Lighthouse runs on a simulated fast machine and doesn’t capture real interaction patterns.
The two tools to use:
Google Search Console → Core Web Vitals report. This uses actual Chrome user data (CrUX) from real visitors. It’s the only number that matches what Google is ranking on. Look at the mobile report specifically.
Chrome DevTools → Performance panel, with CPU throttling set to 4x slowdown. Record a session where you tap three or four things. Look for long tasks on the main thread. Anything over 50ms is a problem; over 200ms is the problem.
Between those two, you’ll know whether INP is a real issue for your site and where the time is going.
The three things that fix it
In the order I try them:
Remove scripts. Open your site’s source and count the third-party <script> tags. For most small business sites, you need analytics and that’s it. The chat widget can load on the contact page only. The remarketing pixel can load on conversion pages only. The A/B testing tool that nobody has looked at in a year can be deleted. Removing five or ten scripts often moves INP more than any other change.
Defer what’s left. Every non-critical script should have defer or async, and ideally load after the main page is interactive. This includes analytics. A delay of a few hundred milliseconds before analytics loads doesn’t change anything that matters and buys you back main-thread time on the first interaction.
Use an islands architecture, or don’t hydrate at all. If you’re on Next.js or similar and shipping a React tree for a mostly-static marketing page, that’s the problem. Astro’s islands model hydrates only the components that actually need JavaScript, which is usually two or three on a marketing site. Switching frameworks is a big lift; if you can’t, at least audit which components are hydrating and why.
What not to do
Don’t install a “speed optimisation” plugin and assume it fixed things. Those plugins mostly add caching, which helps LCP, not INP. INP is a main-thread problem; caching doesn’t solve it.
Don’t buy a faster server. Same reason.
Don’t add an AMP version. AMP is in maintenance mode and Google stopped giving it preferential treatment in 2021.
Why it’s worth the work
The March 2026 core update amplified the gap between sites that pass all three Core Web Vitals and sites that fail any of them. The penalty for a bad INP is no longer mainly about users bouncing — that was already a problem — it’s increasingly about rankings.
For a small business site, that’s usually the difference between showing up on page one for “painter near me” and showing up on page two. Same site, same content, same backlinks. Just faster.
If you want a specific read on your site’s INP and what’s causing it, send us a note.