Page Load Time vs Conversion Rate: What the Data Says

The classic stats get quoted at every conference: a 1-second delay costs 7% of conversions; 53% of mobile visits abandon after 3 seconds; Amazon loses millions per 100 ms. The numbers are dated and the studies imperfect — but two decades of replication agree on the shape: conversion falls as load time rises, the fall is steepest early, and most teams have no idea where on the curve their own pages sit. That last part is the fixable one.

The curve, not the slogan

The relationship is roughly exponential decay, which produces three practical zones:

• Under ~1.5 s: speed is invisible. Going from 1.2 s to 0.9 s is engineering pride, not revenue.
• ~1.5 s to ~4 s: the steep zone. Every 500 ms here measurably moves bounce and conversion. Most real-world money is won or lost in this band.
• Over ~4 s: the survivors zone. Visitors still present are unusually motivated; the casual majority already left.

Corollary: the average load time of your site is nearly useless. What matters is which high-intent pages sit in the steep zone, for which segment of visitors.

Why averages lie about your real exposure

A pricing page averaging 1.8 s sounds fine — until you split it: 1.1 s on desktop fiber, 4.2 s at p75 on mobile. The expensive truth lives in the percentiles and the segments, never the mean. This is the same logic as field-data vitals monitoring: p75, per page, per device class.

Finding your own leaks (the per-visit method)

Aggregate studies tell you the industry curve; your decisions need your curve. The method:

1. Record load time on every visit — not in a separate RUM tool, but on the same record as behavior, so timing and outcome are joinable. Clycyo's tracker captures navigationStart → loadEventEnd per pageview (and frame-time per SPA transition) automatically.
2. Compare converting vs non-converting sessions on the same page. If non-converters consistently saw slower loads, you have a leak with a number on it.
3. Read individual journeys for the mechanism. The aggregate says 'slow pricing page correlates with abandonment'; one journey shows the visitor who waited 5.3 s, scrolled once, and left. That visit is on a real timeline you can open — e.g. in the live demo — and it converts a statistic into a bug report.
4. Price the fix: (visitors in the steep zone) × (conversion gap) × (average order value) = the budget your performance sprint deserves.

The usual suspects, ranked by effort-to-impact

1. Oversized images on landing pages — an afternoon of compression for the single biggest LCP win on most sites.
2. Third-party scripts — every marketing tag is main-thread time; audit which ones still earn their keep (the tracker tax is real and measurable).
3. No CDN, or CDN misses in your second-biggest market.
4. Render-blocking fonts and CSS — font-display: swap remains free money.
5. Slow API calls blocking first render on SPA routes.

The honest summary

You do not need the fastest site on the internet. You need your high-intent pages out of the steep zone for the devices your customers actually use — and the measurement to notice when a deploy pushes them back in. That is a per-visit measurement problem, and it is exactly why Clycyo treats page-load time as a first-class analytics dimension rather than a separate tool's job.