Open your analytics and split conversion rate by device. Almost every site shows the same shape: mobile brings the majority of the visits and a minority of the money, and desktop converts at something like twice the rate.
This has been true for long enough that people have stopped treating it as a problem and started treating it as weather. It is not weather. Some of it is genuinely structural and some of it is your checkout, and the two need separating before you can act on either.
Directional composite of commonly reported e-commerce splits. The exact figures vary widely by sector. The inversion between the first pair and the second is the durable finding.
The shape almost every site sees. Share of visits, mobile: 65%. Share of visits, desktop: 35%. Share of revenue, mobile: 40%. Share of revenue, desktop: 60%.
The part that is genuinely structural
Some of this gap is real behaviour and no redesign will remove it.
Phones are where people browse. Desktops are where people commit. A commute, a sofa, a queue: these are discovery contexts, and a considered purchase often gets deferred to a moment with a bigger screen and a keyboard. That deferral is invisible to your analytics, which sees a mobile session that bounced and a desktop session that converted, and reports them as two unrelated people.
Cross-device attribution is the technical name for the problem, and it is the reason we care about it here. If a mobile ad click leads to a desktop purchase two days later, most measurement setups will credit the desktop channel entirely. Your mobile campaigns then look worse than they are, you cut them, and the desktop conversions they were feeding quietly decline.
The part that is your checkout
The rest of the gap is friction that only exists on small screens, and it is unglamorous work with reliable returns.
- 01Form fields that trigger the wrong keyboard. A numeric field without the numeric inputmode makes every card entry a fight
- 02Broken or missing autofill. Field names browsers cannot recognise turn a 30 second checkout into a two minute one
- 03Tap targets under 44px, or spaced closely enough that thumbs hit the wrong one
- 04Sticky headers and cookie banners eating a third of a small viewport
- 05Layout shift as images and ads load, moving the button as it is being tapped
- 06Payment methods that suit desktops. Apple Pay and Google Pay skip the entire form
The autofill point is worth dwelling on because it is invisible in testing. Developers rarely have realistic saved addresses and cards, so autofill works in the office by not being needed. On a real device with real saved data, a form with unconventional field names silently refuses to fill and the user types 40 characters they should never have typed.
Baymard's checkout research puts the average US checkout at 23.48 form elements against an achievable 12 to 14. Every one of those fields is worse on a phone. That is the connection between the two halves of this piece: the mobile conversion gap is partly a general checkout problem that a small screen amplifies until it becomes visible.
How to tell which half you have
Compare the funnel, not the conversion rate. If mobile and desktop track each other closely until checkout and then diverge sharply, you have a checkout problem and it is fixable this quarter. If mobile falls behind from the very first step, you have a traffic or expectation problem and the checkout is not the place to spend.
For the cross-device half, the honest position is that you cannot fully solve it and you can stop being fooled by it. Attribute a visitor consistently across the whole funnel rather than per step, count the journeys you can see, and state the size of what you cannot. Our piece on why your ad platform and your analytics disagree covers the mechanics, and the average checkout asks 23 questions covers the field-count work directly.