A site that loads in 4 seconds has already crossed into Google's "poor" Largest Contentful Paint band — the point past which most visitors are gone before the page finishes rendering. Conversion drop-off accelerates past 2.5 seconds, mobile users first. Speed isn't a technical nice-to-have; Core Web Vitals gives you the exact numbers to build against.
- Google's Core Web Vitals set real thresholds for Largest Contentful Paint: under 2.5 seconds is good, 2.5–4 seconds needs improvement, above 4 seconds is poor — a 4-second load already sits in the poor band.
- Conversion drop-off isn't linear with load time. It compounds past the 2–3 second mark, and mobile connections hit that wall first.
- INP (Interaction to Next Paint) and CLS (Cumulative Layout Shift) matter as much as raw load time — a page that loads fast but jumps around while doing it still fails the standard.
- Most slow sites aren't slow because the design is heavy. Unoptimized images, unnecessary third-party scripts, and missing caching account for most of it.
Run your homepage through PageSpeed Insights and see a 4-second load time, and it's tempting to file that under "could be faster" and move on. Don't. Four seconds isn't a middling grade — it's the exact point where Google's own Core Web Vitals stop calling your page "needs improvement" and start calling it "poor." Long before Google does anything about that, your visitors already have.
What "4 seconds" actually means
Core Web Vitals grade three things, and load time isn't one of them directly — the closest proxy is Largest Contentful Paint (LCP): how long it takes the biggest visible element on the page to render. Google's own published bands are under 2.5 seconds for good, 2.5 to 4 seconds for needs improvement, and above 4 seconds for poor.
A site loading in 4 seconds isn't in a grey zone. It's at the exact line where Google's own definition of acceptable ends. The other two Core Web Vitals matter just as much: Interaction to Next Paint (INP) under 200 milliseconds is good, and Cumulative Layout Shift (CLS) under 0.1 is good. A page can pass LCP and still fail the standard if buttons take a second to respond or content jumps around while images load in.
Why the drop-off isn't gradual
The relationship between load time and how many visitors stick around doesn't move in a straight line — it compounds. A visitor who waits two seconds is mildly annoyed and still there. A visitor who waits four is often already gone, and mobile visitors feel this first: every extra second on a patchy connection is felt directly, not just measured in a lab tool. This is also where paid traffic gets expensive twice — once to win the click, and again when the landing page loses the visitor before the offer even loads.
Where the four seconds actually goes
Almost none of it is usually a design problem. The usual culprits: unoptimized images shipped at full resolution and resized in the browser, render-blocking scripts loading before anything visible appears, a stack of third-party widgets (chat, extra pixels, unused fonts) each adding their own request, and no caching or CDN in front of the origin server. Fixing these is an engineering pass, not a rebuild — we run this as part of the audit stage in how we work with a new account, before touching a single design decision.
Treat page speed the same way you'd treat a leaking step in any other part of the funnel: it doesn't matter how good the ad, the offer, or the design is if the page itself is the reason people leave before any of that gets a chance to work.
Related service: Web Design & Development