Page Speed and Conversion Statistics
A sourced, dated roundup of how load time affects online store revenue. It covers bounce and conversion by second, Core Web Vitals thresholds and pass rates, and published case studies from Rakuten 24, Swappie, Vodafone, Zalando and more.
Page speed and conversion statistics
Published, sourced figures showing how page load time changes bounce rate, conversion rate and revenue for online stores.
Page speed and conversion statistics measure what each second, or each 100 milliseconds, of load time costs a store. The evidence comes in three types: large observational studies (Google/SOASTA, Akamai, Portent), controlled before-and-after studies (Deloitte's 'Milliseconds Make Millions'), and individual company case studies published by Google on web.dev.
Since 2021 most speed work has been framed around Google's Core Web Vitals: Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS). This page collects the most-cited figures, with each one's source and year, so you can quote them accurately and know how much weight each deserves.
Key statistics
• A 0.1-second improvement in mobile site speed lifted retail conversion rates by 8.4% and average order value by 9.2% (Deloitte & Google, 'Milliseconds Make Millions', 2020). • As mobile load time goes from 1 to 3 seconds, the probability of a bounce rises by 32%; from 1 to 5 seconds, by 90% (Google/SOASTA Research, 2017). • 53% of mobile site visits are abandoned if a page takes longer than 3 seconds to load (Google DoubleClick, 2016). • A 100-millisecond delay in load time can cut conversion rates by 7% (Akamai, State of Online Retail Performance, 2017). • An ecommerce site that loads in 1 second converts 2.5x higher than one that loads in 5 seconds (Portent, 2022). • Between 0 and 5 seconds, conversion rates fall by an average of 4.42% with each extra second of load time (Portent, 2019). • Rakuten 24 saw revenue per visitor rise 53.37% and conversion rate rise 33.13% after Core Web Vitals work (web.dev case study, 2022). • Swappie cut mobile LCP by 55% and grew mobile revenue by 42% (web.dev case study, 2022). • Zalando found that a 100ms improvement in load time raised revenue per session by 0.7% (Zalando engineering, 2018).
Most speed statistics you see quoted are 5-10 years old, and many are copied without their source. Below, each figure has its publisher, year and method, so you can tell a controlled study apart from a vendor survey before you put it in a business case.
Load time vs conversion and bounce: the headline studies
How load time affects conversion and abandonment: most-cited studies
| Finding | Figure | Source (year) | Method |
|---|---|---|---|
| Mobile visits abandoned when load > 3s | 53% | Google DoubleClick (2016) | Publisher site analytics |
| Average mobile site load time on 3G | 19 seconds | Google DoubleClick (2016) | Analysis of 10,000+ mobile domains |
| Conversion change per 100ms delay | -7% | Akamai (2017) | Retail beacon data, billions of visits |
| Bounce rate change from a 2s delay | +103% | Akamai (2017) | Retail beacon data |
| Load time with peak conversion | 1.8-2.7s (by device) | Akamai (2017) | Retail beacon data |
| Conversion drop per extra second (0-5s) | -4.42% avg | Portent (2019) | 100M+ page views, 20 sites |
| Ecommerce conversion, 1s site vs 5s site | 2.5x higher | Portent (2022) | Ecommerce sessions sample |
| Retail conversion lift from 0.1s faster mobile | +8.4% | Deloitte & Google (2020) | 37 brands, ~30M sessions |
| Retail AOV lift from 0.1s faster mobile | +9.2% | Deloitte & Google (2020) | 37 brands, ~30M sessions |
| Travel conversion lift from 0.1s faster mobile | +10.1% | Deloitte & Google (2020) | 37 brands, ~30M sessions |
| Shoppers who say speed affects willingness to buy | ~70% | Unbounce (2019) | Consumer survey |
The Deloitte figure carries the most weight because it tracked the same brands before and after their speed changes, across real sessions. Akamai and Portent show correlation across many sites. They're strong evidence of a direction, not exact predictions for your store.
Note what these studies have in common: the biggest losses come in the first few seconds. Going from 1.5s to 2.5s usually costs more than going from 6s to 7s, because the shoppers who were going to leave have mostly gone already. For a Shopify or WooCommerce store, that means product and collection pages on mobile are where speed work pays back first.
Bounce probability by mobile load time
Increase in bounce probability vs a 1-second page load (mobile)
The curve is steep early and flat later. Most of the extra bounce risk builds between 1 and 5 seconds (+90%), and only 33 more points are added between 5 and 10 seconds. The same 2017 Google/SOASTA analysis found that as page elements grow from 400 to 6,000, the chance of a conversion falls by 95%.
Page weight matters as much as server speed. Each tracking pixel, heatmap recorder and client-side A/B testing script adds requests and main-thread work. Stores running GA4, Hotjar and a separate testing tool often load three overlapping scripts on every page. That's why removing duplicate tools tends to show up as an LCP and INP gain, not just a lower software bill.
Core Web Vitals: thresholds and pass rates for online stores
Core Web Vitals 'good' thresholds and approximate mobile pass rates by platform
| Metric / segment | Good | Needs improvement | Poor | Notes |
|---|---|---|---|---|
| Largest Contentful Paint (LCP) | ≤ 2.5s | 2.5-4.0s | > 4.0s | Usually the hero or product image |
| Interaction to Next Paint (INP) | ≤ 200ms | 200-500ms | > 500ms | Hurt by heavy third-party JavaScript |
| Cumulative Layout Shift (CLS) | ≤ 0.1 | 0.1-0.25 | > 0.25 | Hurt by late-loading banners, reviews and pop-ups |
| Assessment basis | 75th percentile of real-user visits | — | — | Split by mobile and desktop |
| Shopify origins, mobile CWV pass rate | ~65-75% (approx.) | — | — | Hosted, CDN-backed themes |
| WooCommerce origins, mobile CWV pass rate | ~35-45% (approx.) | — | — | Depends heavily on hosting and plugins |
| Magento / Adobe Commerce origins, mobile pass rate | ~25-40% (approx.) | — | — | Often heavy custom themes and extensions |
Platform pass rates mostly reflect who builds on each platform. Shopify controls hosting, the CDN and the checkout, so stores start from a strong baseline and usually fall out of 'good' because of apps and theme add-ons. On WooCommerce and Magento, hosting, caching and extension choices decide the result, so stores on the same platform vary much more.
Passing Core Web Vitals is a minimum, not a goal
A 'good' LCP of 2.4s still sits in the part of the Google/SOASTA curve where bounce risk is rising. Core Web Vitals is a pass/fail check at the 75th percentile, so up to a quarter of your mobile shoppers can have a slower experience while the report stays green. Track the 90th percentile for product and checkout pages on mobile separately.
Published speed case studies from retail and ecommerce
Speed improvements and their reported business results
| Company (sector) | Speed change | Reported result | Year |
|---|---|---|---|
| Rakuten 24 (Japanese online retail) | Core Web Vitals optimisation | +53.37% revenue per visitor, +33.13% conversion rate, +15.20% AOV, -35.12% exit rate | 2022 |
| Swappie (refurbished phones) | -55% LCP, -91% CLS | +42% mobile revenue | 2022 |
| Vodafone Italy (telecom ecommerce) | -31% LCP | +8% sales, +15% lead-to-visit rate, +11% cart-to-visit rate | 2021 |
| Renault (automotive) | -1s LCP | -14 pts bounce rate, +13% conversions | 2022 |
| Agrofy (agri marketplace) | -70% LCP | -76% load abandonment | 2021 |
| redBus (travel booking) | INP improvement | +7% sales | 2023 |
| AliExpress (marketplace) | -36% load time | +10.5% orders, +27% conversion for new customers | 2017 |
| Zalando (fashion) | -100ms load time | +0.7% revenue per session | 2018 |
| Walmart (retail) | -1s load time | Up to +2% conversion | 2012 |
| Amazon (retail) | +100ms latency | -1% sales | 2006 |
Read these as upper bounds. Companies publish the projects that worked, and several of these (Rakuten 24, Vodafone) were A/B tests run on a slow starting point. The most useful number to take away is the size of the effect per 100ms or per second of LCP, because you can apply that to your own traffic.
How to use these numbers
1) In a business case, cite Deloitte (2020) for the effect size and Google/SOASTA (2017) for bounce risk, with their years. 2) Don't multiply Akamai's 7% per 100ms across several seconds; the effect isn't linear. 3) Use your own CrUX or real-user data for the baseline: mobile LCP at the 75th percentile on product pages. 4) Turn a speed project into a forecast with a conservative lift, such as 1-3% conversion per 0.5s of LCP saved. Then confirm it with a controlled test, not a before/after comparison that seasonality can skew.
What these figures mean for a store doing €1M-€15M
Take an apparel store on Shopify with €6M revenue, 70% mobile traffic and a 2.0% mobile conversion rate. A conservative 2% relative lift on mobile conversion is worth roughly €84,000 a year. That's well short of Deloitte's 8.4%, and often more than the store pays for its whole analytics and testing stack.
Where to start: compress and correctly size product images (usually the LCP element), reserve space for review widgets and promo bars (CLS), and audit third-party scripts (INP). For most stores, removing one overlapping tracking or testing script is the cheapest speed gain available. Speed sits inside broader ecommerce CRO work, and it improves every other test you run.
Page speed and conversion: frequently asked questions
It depends on your starting point. Portent (2019) measured an average 4.42% drop in conversion per extra second between 0 and 5 seconds. Older claims of a '7% loss per second' come from a misreading of Akamai's 2017 figure, which was 7% per 100 milliseconds on retail sites.
Yes. It comes from 'Milliseconds Make Millions' by Deloitte with Google (2020), covering 37 brands and about 30 million sessions. A 0.1s mobile speed improvement was linked to an 8.4% higher retail conversion rate and 9.2% higher average order value. The effect is real, but don't expect it to repeat for every 0.1s you save.
Aim for mobile LCP under 2.5 seconds at the 75th percentile, Google's 'good' threshold, and ideally under 2 seconds on product pages. Akamai's 2017 retail data found conversion peaked at load times of 1.8-2.7 seconds depending on device.
Yes, but only a little. Core Web Vitals are part of Google's page experience signals, and relevance and content quality matter more. The commercial case for speed comes mainly from conversion and bounce, not rankings.
Most Shopify stores do. Shopify origins have among the highest mobile pass rates in the HTTP Archive Core Web Vitals Technology Report. Stores usually fail because of installed apps, heavy theme sections, unoptimised hero images and third-party scripts, not because of the platform.
Those platforms are self-hosted. Hosting quality, caching, database setup and plugins all vary from store to store. A well-tuned WooCommerce store on good hosting can match Shopify, but the average one doesn't.
They can. Each script adds network requests and main-thread work, which shows up in LCP and INP. Client-side testing tools can also cause flicker or delay rendering. Running several overlapping tools (analytics, session recording, testing) multiplies the cost, so combining them into one lightweight snippet is a common speed fix.
For most stores, LCP on mobile product and collection pages, because it controls when shoppers see the product. INP becomes critical on pages with variant pickers, filters and cart drawers. CLS mostly affects trust and mis-taps around buy buttons.
Run a controlled test where possible, such as a split test of an optimised template or a holdout on part of your traffic. Before/after comparisons get skewed by seasonality and campaigns. Track conversion rate, revenue per visitor and bounce by device, alongside real-user LCP and INP.
As a direction, yes. As precise figures, not really. Amazon's '100ms = 1% of sales' (2006) and Walmart's '1s = up to 2% conversion' (2012) predate modern mobile commerce. Prefer the Deloitte (2020) study and recent web.dev case studies like Rakuten 24 and Swappie when you quote figures today.
Find what's costing your store conversions — then test the fix
Metricuno shows where shoppers drop off, lets you change the page without a developer, and runs the A/B test that proves it. Free for 30 days. No credit card.