← Back to the blog

Site Speed & Revenue-Loss Audit

Does Website Speed Affect Conversions? What the Research Shows

By Hamza IqbalFounder, ToolsForge

Reviewed 5 min readHow we source numbers

Stopwatch beside a shopping cart and checkout button

Does website speed really affect conversion rate?

Yes. In Portent's April 2022 analysis of about 100 million page views across 20 sites, online-store pages that loaded in one second converted at 3.05%, falling to 1.68% at two seconds and 1.12% at three. Deloitte's 2020 study found a 0.1-second faster mobile site lifted retail conversions by 8.4%.

The decline starts at the first second. There's no load time below which speed stops mattering.

What do the studies actually show?

Study What was measured Result
Portent, April 2022 About 100 million page views, 14 B2B lead-gen and 6 B2C store sites Store conversion 3.05% at 1s, 1.68% at 2s, 1.12% at 3s. B2B lead-gen conversion almost 40% at 1s, 34% at 2s, 29% at 3s
Deloitte, Milliseconds Make Millions, March 2020 Mobile sites of retail, travel, luxury and lead-gen brands in Europe and the US, over four weeks A 0.1s improvement lifted retail conversions 8.4% and retail order value 9.2%; travel conversions rose 10.1%
Vodafone A/B test, web.dev, March 2021 Two versions of a landing page, paid traffic split 50/50 The version with 31% better LCP made 8% more sales, with 15% better lead-to-visit and 11% better cart-to-visit rates

Two cautions. Portent and Deloitte are correlation studies across many sites, so they show a strong pattern, not a guarantee for yours. Vodafone's is a controlled test on one page, which is stronger evidence but a single case. You'll also see the claim that every 100ms cost Amazon 1% of sales; it comes from a former engineer's description of internal tests in 2006 and was never published as a study.

Portent's B2B and store columns measure different things: a B2B "goal" is a form fill or download, which converts far more often than a purchase. Its B2B one-second pages converted three times as well as five-second pages and five times as well as ten-second pages.

Why does a few seconds of delay change buying behavior?

  • People leave before your offer loads. A headline, price or trust badge can't persuade anyone who has already gone back to the search results.
  • Slow reads as unreliable. A page that stalls or jumps while loading makes first-time visitors less willing to hand over card details.
  • Mobile makes it worse. Phones on cellular connections are slower and less consistent than the office wifi a site was built on, so a page that feels fine in development can be slow for real visitors.
  • The checkout is usually slow too. If the server and assets are slow on the first page, they're slow on the payment step, where friction costs the most.

Is this just about e-commerce?

No. The research leans on stores because purchases are easy to count, but Portent's B2B lead-gen sites showed the same slide, and Deloitte measured lead-gen pages as well. Any page that asks for an action (a quote request, a booking, a trial signup) loses impatient visitors the same way. For a service business the cost can be higher per visitor, because one lost lead may be worth more than one lost cart, and there's no abandoned-cart email to win it back.

How do you estimate what slow pages cost your own site?

You need three numbers: your current load time (LCP is the best proxy), monthly visitors to the pages that sell, and your conversion rate and average order value.

Worked example. Assume a store gets 20,000 product-page visits a month, converts at 2.5% and has a $60 average order, all example figures:

Step Result
Orders a month (20,000 × 2.5%) 500
Revenue a month (500 × $60) $30,000
Real-visitor LCP 4.0s, which is 1.5s over Google's 2.5s "good" threshold
Estimated loss, using Portent's 2019 rate of 4.42% fewer conversions per extra second (1.5 × 4.42%) About 6.6% of conversions
In money (6.6% × $30,000) About $2,000 a month, roughly 33 orders

That is the method our site speed audit uses, with the real LCP measured for you. It uses Portent's earlier 2019 study, which reported a per-second rate (an average 4.42% drop per extra second for pages loading in 0–5 seconds); the 2022 update on the same Portent URL reports conversion rates by load time instead, which is what the table above quotes. Treat the result as an order-of-magnitude range, not a forecast. Don't scale Deloitte's 0.1-second figure up to a full second: Deloitte didn't measure that, and the effect won't be linear.

Is speed a one-time fix?

Usually not. Sites slow down as plugins, tracking scripts, chat widgets and larger images pile up. A site that was fast at launch can be back in the slow range a year or two later. Re-test every quarter, and after any redesign or new plugin, the same way you'd review analytics.

What counts as fast enough?

Aim for a Largest Contentful Paint of 2.5 seconds or less for real visitors (Google's "good" threshold). In Portent's 2022 data the highest conversion rates were at one second, so under 2.5s is a floor, not a finish line. Why is my website so slow? covers the usual causes, and Core Web Vitals explained covers the metrics. If you're choosing a testing tool, see PageSpeed Insights vs GTmetrix vs Pingdom.

FAQ

Usually, yes. Phones on cellular networks load pages more slowly and less predictably, and Deloitte's 8.4% retail lift came from mobile sites specifically. Test mobile first.
It depends on where you start. Deloitte measured an 8.4% retail conversion lift from a 0.1-second mobile improvement, and Vodafone's A/B test found 8% more sales from a 31% better LCP. Gains are largest on pages that are currently slow.
Multiply your monthly revenue by an estimated conversion loss. Using Portent's 2019 rate of about 4.42% per second beyond a 2.5s LCP, a $30,000-a-month store at 4.0s would lose around 6.6%, or about $2,000 a month. It's a rough range, not a guarantee.
Yes, but a modest one. Google uses real-user Core Web Vitals in its ranking systems, and relevance comes first. Does page speed affect SEO? covers what Google has said.

Site Speed & Revenue-Loss Audit

Find out what slow load times are costing you.

Run the free tool

Written by

Hamza Iqbal

Founder, ToolsForge

Founder of ToolsForge and a WordPress & WooCommerce developer. Builds the free calculators on this site and writes the guides behind them — every guide pairs real, sourced numbers with a tool so you can run your own.