More sales from the traffic you already have
Mobile carries three quarters of ecommerce traffic and, depending on whose panel you read, converts either 37% worse than desktop or slightly better. Both findings are real. I lay out the published numbers, then add a live CrUX pull that explains part of the contradiction.
If you read only one benchmark provider, you will come away certain about the device gap. Read two and the certainty collapses in an interesting way. The small-store panel says phones convert far worse than desktops. The enterprise panel says the gap is gone, even slightly reversed. This entry puts both on the record, then adds a measurement of my own that explains part of the contradiction.
Littledata, 2,800 sites, heavily Shopify and mostly small to mid-size: mobile converts at 1.2% against desktop’s 1.9%. Checkout completion shows the same slope, 44% mobile against 49% desktop. That is the classic gap: a desktop visitor is roughly 1.6 times likelier to end up buying.
Dynamic Yield, 400+ mostly enterprise brands, twelve months to June 2026: mobile 2.86%, desktop 2.46%, tablet 2.89%. On this panel the phone is not the weak device; it is marginally the strong one.
Meanwhile the traffic split is not in dispute anywhere: Dynamic Yield’s panel puts mobile at roughly 76% of visits over the trailing year, desktop at 23%, tablet at a rounding error. Whatever converts on phones is what converts, full stop.
I read the contradiction as the most useful benchmark on this page. The enterprise brands in Dynamic Yield’s panel have spent a decade turning phones into first-class buying devices: express payment buttons that skip the address form entirely, product pages built thumb-first, apps and saved cards catching returning buyers. The median store in Littledata’s panel has a responsive theme and calls it a day. The device gap, in other words, is not a law of nature about phones. It is a measurement of how much work a store has put into mobile, which means it is closable, and the enterprise panel is the proof.
Part of the gap is also pure physics, and this part I can measure myself rather than take on faith.
I pulled Chrome UX Report field data for six well-known fashion storefronts, phone versus desktop, for the 28-day window ending July 15, 2026. LCP p75, real users, same origins:
| Store | Mobile LCP p75 | Desktop LCP p75 | Mobile penalty |
|---|---|---|---|
| Gymshark | 2.24s | 1.45s | +0.79s |
| Oh Polly | 2.21s | 1.86s | +0.35s |
| Hello Molly | 2.03s | 1.80s | +0.23s |
| Allbirds | 1.44s | 1.32s | +0.12s |
| Fashion Nova | 2.34s | 2.34s | 0.00s |
| Meshki | 2.12s | 2.16s | -0.04s |
Source: CrUX API, June 18 to July 15, 2026, p75 per form factor.
Five of six origins are slower in the hands of real phone users, by up to eight tenths of a second. Run that through the published speed studies, Renault’s 13% of conversions per second of LCP, and the physics alone predicts phones should convert somewhat worse everywhere. That the enterprise panel still shows parity tells you how much the express-payment and design work compensates. The lesson cuts both ways: a small store gets the mobile physics penalty and the mobile design penalty, stacked.
Each row is one POST request per form factor to the Chrome UX Report API (chromeuxreport.googleapis.com/v1/records:queryRecord), querying the origin with formFactor PHONE and then DESKTOP, in July 2026. Values are 75th percentile Largest Contentful Paint over CrUX’s rolling 28-day window, June 18 to July 15, 2026, covering real Chrome users only. Origins were chosen as recognizable fashion storefronts with sufficient CrUX traffic; two candidates I tried (Princess Polly, Shein’s US origin) returned no origin-level record and were excluded rather than substituted with lab data. CrUX reports experience per origin, not per device model, so this measures each store’s delivered speed to its actual phone audience, network conditions included. The conversion figures in this entry are from Littledata and Dynamic Yield as cited, not from me.
First, I compute the store’s own gap honestly: mobile conversion divided by desktop conversion, from Shopify Analytics, over at least a month. Against Littledata’s averages, 1.2 over 1.9 makes the benchmark ratio 0.63. A store at 0.8 or better is ahead of the median; a store at 0.4 has a mobile-specific fire.
Second, when the ratio is bad, I audit only on a phone until I understand why, the habit I defended in why I audit mobile first. The usual finds: express payments switched off, a cookie banner eating the first screen, form fields that trigger the wrong keyboard, and the speed penalty measured above.
Phones stopped being the future of buying years ago. They are the present that half the stores still haven’t moved into.
Third, I resist the fashionable conclusion that desktop no longer matters. On the fashion panel above, desktop is a quarter of sessions, but on high-ticket stores it routinely carries most of the revenue per session. The device gap is a ratio to manage, not a war to pick a side in.
On the average small store, yes: Littledata's Shopify benchmark has mobile at 1.2% against desktop at 1.9%, and mobile checkout completion at 44% against 49% on desktop. But Dynamic Yield's large-brand panel shows mobile at 2.86% slightly ahead of desktop at 2.46% over the twelve months to June 2026, so the gap is a property of stores, not of phones.
Three compounding reasons. Intent mix: phones host more idle browsing, so more sessions were never going to buy. Physics: mobile networks and CPUs make the same store slower on a phone, which the speed studies price at roughly 8 to 13% of conversions per second. And friction: typing addresses and card numbers with a thumb punishes every unnecessary field twice as hard.
Audit mobile first, always: it is where roughly 76% of traffic lives per Dynamic Yield's panel, and where the defects are worst. But check your own revenue split by device before reallocating design effort, because on high-ticket stores desktop often still carries the majority of the money on a minority of the sessions.
Against Littledata's Shopify panel, 1.2% is average for mobile and above 3.9% puts you in the best 10%. If your mobile rate is within about 20% of your desktop rate, you are already doing better than the median store in that benchmark.