Does LCP Affect Conversion Rate? What the Evidence Shows

Yes — LCP affects conversion, because Largest Contentful Paint is roughly when your main content becomes visible, and a visitor can't act on a value proposition or a CTA they can't see yet. Google's web.dev defines a "good" LCP as 2.5 seconds or less. Rather than lean on someone else's headline statistic, the honest way to prove the link for your site is to measure LCP per session and compare conversion across the speed bands.

The mechanism, not a borrowed stat

It's easy to quote a dramatic "every second costs X% of conversions" figure, but those numbers come from specific sites with specific traffic and don't transfer. The mechanism, though, is general and uncontroversial:

• LCP marks when the largest above-the-fold element — usually your hero or headline — finishes painting.
• Until then, the visitor sees a blank or partial page. They can't read your offer or click anything.
• The longer that lasts, the more visitors give up before the page is usable, and the later everyone else can engage.

So LCP doesn't "cause" conversion directly; it gates the moment conversion can begin. That's enough to make it worth fixing.

The thresholds that matter

web.dev's Core Web Vitals bands for LCP, measured at the 75th percentile of real users:

• Good: ≤ 2.5 s
• Needs improvement: 2.5 s – 4 s
• Poor: > 4 s

The goal is to get most of your sessions into the good band — not to chase a perfect lab score.

Measure LCP for real users in PostHog

Capture LCP in the field with the web-vitals library and send it as an event:

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

Then bucket your real-user LCP into the web.dev bands:

SELECT
  multiIf(
    toFloat(properties.value) <= 2500, 'good',
    toFloat(properties.value) <= 4000, 'needs improvement',
    'poor'
  ) AS lcp_band,
  countDistinct(person_id) AS sessions
FROM events
WHERE event = 'web_vitals'
  AND properties.metric = 'LCP'
  AND timestamp > now() - INTERVAL 30 DAY
GROUP BY lcp_band
ORDER BY lcp_band

Illustrative sample output:

lcp_band	sessions
good	4,120
needs improvement	1,180
poor	640

Prove the conversion link on your own data

The distribution above tells you how fast your site is; to show LCP affects your conversion, compare the conversion rate of fast versus slow sessions. Capture both web_vitals and signup_completed, then look at whether the good-LCP cohort converts better than the poor-LCP cohort. If it does — and it usually does — you've measured the impact directly instead of borrowing a number.

What to do about it

If a meaningful share of sessions land in needs-improvement or poor, the highest-leverage fixes are almost always the hero image (preload it, don't lazy-load it) and render-blocking resources. See the framework-specific guides for the exact changes. If you'd like the slowest, lowest-converting part of your page found and fixed as a Pull Request each week, that's what Velyr does.