Does INP Affect Conversion on a Landing Page?

Yes — INP affects conversion, because Interaction to Next Paint measures how quickly your page responds to a tap or click, and a CTA that feels laggy reads as broken. A visitor clicks, nothing happens for a beat, and they either tap again or assume it's stuck. Google's web.dev defines a good INP as 200ms or less. INP replaced First Input Delay as a Core Web Vital in 2024.

The mechanism

INP captures the delay between an interaction and the next visible paint — essentially, "when I clicked, how long until the page reacted?" It hurts conversion because responsiveness is how reliable a page feels:

• A CTA that responds instantly feels solid; the visitor trusts the click landed.
• A CTA that hangs for half a second after the click feels broken. The visitor double-taps (risking a double-submit) or gives up.

On a landing page, this most often bites at the exact worst moment — the click on the primary CTA — because that's when any heavy JavaScript fires.

The thresholds

web.dev's INP bands at the 75th percentile of real users:

• Good: ≤ 200 ms
• Needs improvement: 200 – 500 ms
• Poor: > 500 ms

What causes poor INP

The root cause is the main thread being busy when the user interacts:

• Heavy event handlers — a click that triggers an expensive calculation or a large re-render.
• Too much JavaScript — large bundles that keep the main thread busy parsing and executing.
• Third-party scripts — analytics, chat widgets, and tag managers competing for the main thread.

The fixes are to ship less JavaScript, defer what isn't needed for the first interaction, and break long tasks into smaller chunks.

Measure your field INP in PostHog

Capture INP with the web-vitals library and bucket it:

import { onINP } from 'web-vitals'
onINP((metric) => {
  posthog.capture('web_vitals', { metric: 'INP', value: metric.value })
})

SELECT
  multiIf(
    toFloat(properties.value) <= 200, 'good',
    toFloat(properties.value) <= 500, 'needs improvement',
    'poor'
  ) AS inp_band,
  countDistinct(person_id) AS sessions
FROM events
WHERE event = 'web_vitals'
  AND properties.metric = 'INP'
  AND timestamp > now() - INTERVAL 30 DAY
GROUP BY inp_band
ORDER BY inp_band

Illustrative sample output:

inp_band	sessions
good	4,300
needs improvement	760
poor	220

If sessions are slipping into needs-improvement or poor, your CTA may feel sluggish exactly when it matters.

Reduce the work

Defer non-critical scripts so they don't compete with the first interaction, trim your bundle, and keep click handlers light. On a landing page, most of the third-party JavaScript can load after the page is interactive. If you'd like the heaviest, most conversion-critical interaction found and fixed as a Pull Request, that's what Velyr does.