Performance glossaryFirst Input Delay (FID) measured the delay before a page could respond to its first interaction. It is retired: Interaction to Next Paint (INP) replaced it as a Core Web Vital in March 2024.
First Input Delay (FID) was one of Google's original three Core Web Vitals. It measured responsiveness: how long a page made a visitor wait before it could start handling their first click, tap, or key press. In March 2024, Interaction to Next Paint (INP) replaced it, and Google Search Console, PageSpeed Insights, and the Chrome UX Report now report INP instead.
FID still turns up in older reports, plugin settings, and articles, so it helps to know what it measured and why it was retired.
What FID measured
When a visitor interacts with a page, the browser has to run the page's event handlers on its main thread. If the main thread is busy, usually running JavaScript while the page loads, the interaction waits in a queue. FID was the length of that wait, for the first interaction only:
- It counted discrete actions such as clicks, taps, and key presses, not scrolling or zooming.
- It measured only the input delay, not the time to run the event handlers or to paint the result.
- It needed a real person to interact, so it could only be measured in the field, not in a lab test.
A good FID was 100 milliseconds or less, and over 300 milliseconds was poor.
Why INP replaced it
FID had two blind spots: it only looked at the first interaction, and it only measured part of that interaction. A page could respond quickly to the first click, then freeze for half a second every time someone opened a menu, and still pass. Most sites passed FID on most pages, so it said little about how responsive pages really felt.
INP fixes both problems. It observes every interaction during the visit and measures the full time from input to the next frame painted, including handler processing and rendering. A good INP is 200 milliseconds or less.
FID, TTI, and TBT
FID was often confused with Time to Interactive (TTI), a lab metric for when a page looks ready to respond reliably. Because lab tools such as Lighthouse could not measure FID, they used Total Blocking Time as a proxy, and TBT remains the lab proxy for INP.
What to do now
If an older report or tool still mentions FID, look at INP instead. The fixes come from the same family, applied more thoroughly: run less JavaScript, defer what is not needed straight away, cut third-party scripts, and break long tasks into smaller ones so the main thread is free to respond.
Source: web.dev: First Input Delay.