More sales from the traffic you already have
Renault, Vodafone and a 30 million session Deloitte study all priced the cost of a slow page. Then I pulled this month's real Chrome field data for six famous stores to see who actually pays it.
Speed is the one conversion factor where I never have to argue from taste. There are real published studies with real sample sizes, and there is a public dataset, Google’s Chrome UX Report, that lets me look up the actual field performance of nearly any store on earth. So this benchmark entry does both: first the published record on what speed is worth, then a fresh CrUX pull I ran this week on six stores everyone knows, to show where the industry’s famous names actually sit.
Renault, 2021. The car maker instrumented its European sites with the web-vitals library and analyzed 10 million visits across 33 countries over four months. Their regression tied a 1 second improvement in LCP to a 13% increase in conversions and a 14 percentage point drop in bounce rate. This is correlation across a huge sample, not a controlled test, but the curve was consistent across markets.
Vodafone, 2021. The cleanest design in the whole literature, because it was a true A/B test. Vodafone Italy served two versions of the same landing page, identical except for performance optimizations, to about 34,000 visits a day. The version with 31% better LCP produced 8% more total sales, an 11% better cart to visit rate, and 15% more leads per visit. Same page, same traffic, same offer. Only the speed differed.
Deloitte and 55, 2020. The Milliseconds Make Millions study analyzed 37 European and American brand sites and over 30 million sessions. Their headline: a 0.1 second improvement in mobile speed metrics was associated with 8.4% more retail conversions and 9.2% higher average order values. For travel sites the conversion lift was 10.1%.
Three methodologies, three industries, one direction. My rule of thumb from these datasets: for a store in the normal speed range, every second of mobile LCP is worth somewhere between high single digits and low teens of relative conversion. Slow pages are not a cosmetic issue, they are a standing tax.
The studies say speed pays. The interesting question for a diary is who is actually collecting. CrUX aggregates real Chrome users’ experiences per origin over a rolling 28 day window, and the API is free, so I queried six well known DTC storefronts for mobile field data. These are 75th percentile values, meaning the experience the slower quarter of real visitors gets, for the collection period of June 16 to July 13, 2026.
| Store | LCP p75 (mobile) | INP p75 | CLS p75 | Verdict vs Google’s 2.5s line |
|---|---|---|---|---|
| Allbirds | 1.4s | 162ms | 0.00 | Well inside, excellent |
| Ruggable | 2.3s | 210ms | 0.04 | Inside, close to the line |
| Gymshark | 2.3s | 388ms | 0.08 | Inside on LCP, INP is the sore spot |
| Fashion Nova | 2.3s | 294ms | 0.19 | Inside on LCP, layout shifts hurt |
| MVMT | 3.0s | 204ms | 0.25 | Outside, LCP and CLS both poor |
| Brooklinen | 3.8s | 304ms | 0.20 | Far outside, slowest of the six |
Source: Chrome UX Report API, Google, July 2026, mobile form factor.
Two things strike me every time I run one of these pulls. First, the spread. These are all successful, heavily invested brands, and yet the gap between the fastest and slowest main-content paint is 2.4 seconds, which the Renault curve would price at something like a quarter of relative conversion. Second, the shape of the failures differs. Gymshark’s LCP is fine but its 388ms INP means the page feels sticky when tapped. MVMT and Fashion Nova ship visible layout shift, the thing that moves the button just as your thumb arrives. Speed is not one number, and neither is the fix.
When I trace a slow ecommerce LCP, the same suspects appear in roughly the same order. Oversized hero imagery shipped at desktop resolution to phones. Third party scripts, the pixel and app collection that grows by one every quarter and never shrinks. Render blocking code from features nobody audited. And underneath everything, the platform and theme baseline the store started from.
That baseline deserves its own line item. Littledata’s platform benchmark measured average server response at 380ms on Shopify against 820ms on Magento and 780ms on WooCommerce, so the platform choice alone moves the floor by nearly half a second before any storefront code runs. On top of that floor, the theme decides how much work every page does. Themes will not save a store that uploads 4MB heroes, but they decide the starting line: Storefront Daily’s Shopify Theme Speed Report 2026 measured real LCP differences between official Theme Store themes, which is worth a look before blaming everything on apps, because a heavy theme is a tax you pay on every page forever.
Every store is running a speed A/B test whether it wants to or not. The slow variant is just called Monday.
The practical routine, since this is the benchmarks section. Once a quarter, query the CrUX API for your origin and for three or four competitors, mobile form factor, and log LCP, INP and CLS at p75. It is one POST request per origin. Plot the quarter over quarter trend next to your mobile conversion rate. You will not get a controlled experiment out of it, but you will get the two things benchmarks are for: an early warning when a release quietly made you slower, and an honest answer to whether your store is fast, not in a lab report, but in the hands of the actual slower quarter of your visitors, the ones the published studies say are deciding whether you get paid.
The published numbers cluster tightly. Renault measured 13% more conversions per 1 second of LCP improvement across 10 million visits. Vodafone's A/B test turned a 31% LCP improvement into 8% more sales. The Deloitte and 55 study of 37 sites associated a 0.1 second improvement with 8.4% more retail conversions. Direction is unanimous, and the size is material.
Google's threshold for a good Largest Contentful Paint is 2.5 seconds or less at the 75th percentile of real users. In my July 2026 CrUX pull, Allbirds sits at 1.4s, Ruggable, Gymshark and Fashion Nova cluster just inside the line at 2.3s, and MVMT and Brooklinen sit outside it at 3.0s and 3.8s.
The Chrome UX Report is Google's public dataset of real Chrome user experiences, aggregated per origin over a rolling 28 days. You can query any origin with sufficient traffic through the free CrUX API or see it inside PageSpeed Insights, which makes it the cheapest competitive benchmark in ecommerce.
Partly. The theme sets the baseline for how much code and layout work every page carries, and Storefront Daily's 2026 measurements of Theme Store themes found meaningful LCP differences between them. Apps, image discipline, and third party scripts then build on or destroy that baseline, which is why identical themes perform differently across stores.