Does site speed affect conversion, and by how much?
Short answer: Yes — but site speed is a hygiene lever, not a growth lever, and almost all the damage sits at the slow end. Directionally, conversion can run several times higher on a page that loads in one second than on one that takes ten, and bounce probability rises sharply between one and three seconds. Below roughly two seconds, extra speed barely moves the number. Speed cannot rescue a weak offer, a wrong price, or copy aimed at nothing.
What the numbers actually say
Across the widely-cited studies, the pattern is consistent even though the exact figures differ by dataset: slow pages leak buyers, and the leak is steepest in the first few seconds. Treat the numbers below as directional industry aggregates, not a forecast for your shop — your own analytics are the only figures that decide anything.
Where the gains actually are
The mistake is treating speed as a dial you keep turning. It is not. The return on speed is front-loaded: moving a five-second page to two seconds recovers real buyers, while shaving a fast 1.4-second page to 1.1 seconds recovers almost none. Speed behaves like a threshold — once a page is comfortably under a couple of seconds, it stops being the constraint, and effort spent there is effort not spent on the actual leak.
This is why "make the site faster" is rarely the right brief. "Find the pages that are slow enough to be losing buyers, and fix those" is. The first is an open-ended engineering project; the second is a diagnosis with an end.
When speed is not your leak
Speed is the wrong lever when your pages already load quickly and buyers still leave. A checkout that loads in one second but asks for an account, a delivery promise that loses against a rival, a price the market rejects, or a product page that never answers the buyer's real risk question — none of these are fixed by more speed. A fast page pointed at the wrong objection converts no better than a slow one; it just fails faster.
Speed sits in the same family as the other non-content levers: it is operations and friction, not copy. It belongs on the list you rule out before you decide the problem is what the page says.
How to tell if speed is your problem
You find out from your own analytics, not from a benchmark. Segment conversion and bounce by device and by page load time: if mobile bounce spikes on your slowest templates while desktop holds, speed is a live suspect. If bounce and conversion look the same across fast and slow pages, it is not — the leak is elsewhere, and a speed project would move nothing. An agent that reads the analytics export can locate that split before anyone opens a performance tool.
What this does not tell you
These figures are aggregates across many shops, industries and years, so the magnitude on your catalogue could be larger or smaller — the only honest number is the one your analytics produce. They are also correlational: fast sites tend to be well-run sites, so some of the measured lift belongs to everything else those teams do well, not to speed alone. And this says nothing about which fix earns its cost; a two-second gain worth chasing on a high-traffic template may be worthless on a page almost no one visits.
FAQ
Does site speed affect conversion, and by how much?
Yes, but it is a hygiene lever and the damage sits at the slow end. Directionally, conversion can run several times higher at a one-second load than at ten seconds, and mobile bounce probability rises about a third between one and three seconds. Below roughly two seconds, extra speed barely moves the number.
What is a good page load time for a webshop?
Comfortably under about two seconds is the practical target; the returns above that are steep and below it flatten quickly. Chasing sub-second loads rarely pays back unless a page is high-traffic and everything else is already optimised.
When is speed not the problem?
When your pages already load quickly and buyers still leave. Then the leak is in price, delivery, checkout friction or copy aimed at the wrong objection — and a speed project moves nothing. Segment your analytics by device and load time before committing.
How do I know if speed is costing me conversions?
Compare conversion and bounce on your fastest versus slowest templates, split by device. A mobile bounce spike on slow templates points to speed; a flat pattern rules it out. Your own data decides, not an industry average.
Directional industry aggregates as cited (Portent; Google / Think with Google; Akamai), not benchmarks for any single shop. Diagnosis method as used in the Cocoon Productions E-Commerce Agent Workflow: segment the analytics export, locate the leak, then choose the lever. Figures directional.
