INP thresholds
| Rating | INP |
|---|---|
| Good | 200 milliseconds or less |
| Needs improvement | Over 200 and up to 500 milliseconds |
| Poor | Over 500 milliseconds |
Like the other Core Web Vitals, web.dev measures INP at the 75th percentile of real page loads, split by mobile and desktop. INP sits next to LCP for loading and CLS for visual stability.
What counts as an interaction
web.dev counts three kinds of input, and only these three:
- Mouse clicks
- Taps on a touchscreen
- Key presses, on a physical or on-screen keyboard
INP watches all of these through the whole visit, not just the first one, and reports one value for the page. One slow button can drag it down, even if the rest of the page feels quick.
INP covers the full span from the input to the next frame the visitor sees. On a phone, a slow INP feels like tapping "Get a quote" and nothing happening, so people tap again or leave.
What slows INP down
web.dev names the main causes: long tasks that keep the browser's main thread busy, heavy work inside event handlers, and costly layout and paint work before the next frame. Its fixes:
- Break up long tasks into smaller pieces.
- Move JavaScript work off the main thread where you can.
- Keep layout and styles simple.
INP is measured from real visits, so a fast office computer can hide the problem. Test on a mid-range phone, tap the menu, tap into the form, and press submit. If anything hangs, that is where to look.
On landing pages, a lot of that main thread work comes from scripts: chat widgets, call tracking, heat maps, and pixels. A free ClickMend scan names unused JavaScript and other slow parts on your ad's landing page, so you can see what to cut. If you use a number swap script, see CallRail and Google Ads landing pages.
ClickMend counts Core Web Vitals as 25% of its score, using real-user CrUX field data when Google has it. When Google does not have enough real visits for a page, the scan uses lab data instead. Either way, the list of slow scripts tells you where to start.
Frequently asked questions
Did INP replace FID?
Yes. web.dev says INP replaced First Input Delay and became a stable Core Web Vital in 2024. FID only looked at the first input. INP looks at all of them.
Does scrolling count toward INP?
No. web.dev says hovering, scrolling, and zooming are not counted, though a click, tap, or key press inside those gestures can be.