The mobile conversion gap is mostly a checkout problem, not a mobile problem

The gap, in numbers

Take a store with 200,000 sessions a month. Seventy percent are mobile (140,000) and 30% are desktop (60,000). Mobile converts at 1.4% and desktop at 3.2%. Average order value is $68 on mobile and $92 on desktop. Here is what that produces:

• Mobile: 1,960 orders, $133,280 revenue, $0.95 per session.
• Desktop: 1,920 orders, $176,640 revenue, $2.94 per session.

Mobile brings more than twice the traffic and fewer orders. The conversion gap is 2.3x, but the revenue-per-session gap is 3.1x, because AOV is also lower on mobile. Revenue per session is the metric that combines both, and it's the one to manage, since optimizing mobile conversion by trading it for a lower basket is not a win.

Not all of the gap is fixable

Before assigning a fix, accept that part of the gap is structural. Mobile traffic skews toward the top of the funnel: social scrolling, email taps, a quick look on the commute. A visitor who opens an ad on their phone at lunch and buys on a laptop that evening is counted as a mobile non-converter and a desktop converter, or in the worst case a direct visit with no source at all.

Your analytics rarely show this cleanly, but you can approximate it. Look at desktop and direct orders from customers whose first recorded session was on mobile, and at the lag between the first mobile touch and the purchase. If, say, 18% of desktop orders began on a phone, then roughly 350 of those 1,920 desktop orders are mobile-assisted, and mobile's true contribution is meaningfully higher than its 1.4% suggests. That doesn't mean mobile needs no work. It means the realistic target is closing part of the gap, not all of it, and that cutting mobile spend based on last-click conversion would be a mistake.

A mobile conversion rate tells you how often a phone completes the sale. It does not tell you how often a phone starts one.

Decompose the gap by funnel step

The useful move is to split overall conversion into its stages and compare devices at each one. Suppose add-to-cart rate is 5.5% on mobile and 7.0% on desktop. Then checkout completion, the share of carts that become orders, is 1.4 / 5.5 = 25.5% on mobile and 3.2 / 7.0 = 45.7% on desktop.

Look at where the gap sits. At the product-page step mobile performs at 79% of desktop (5.5 versus 7.0). At the checkout step it performs at 56% (25.5% versus 45.7%). The cart-to-order step is where the larger loss is, almost twice as bad as the product page. A team that spends the quarter redesigning mobile product pages is working on the smaller half of the problem. This is the same logic as any cart and checkout abandonment diagnostic, with device as the segment that tells you where to point it.

Why mobile checkout loses

Mobile checkout fails for mechanical reasons, and most are fixable without a redesign:

• Typing. Entering a card number, address, email, and phone number on a phone is high effort. Every field is a place to quit. Express wallets such as Apple Pay, Google Pay, and Shop Pay replace all of it with a biometric tap.
• Surprise at the last step. Shipping cost and tax revealed only at checkout hurt on any device, but a phone makes it harder to compare or reconsider, so people just leave.
• Form errors. Wrong keyboard types, address autocomplete that fails, validation that clears the field. These are invisible in aggregate analytics and obvious in session recordings.
• Interrupted sessions. A phone call, a notification, a switch to another app. The cart is abandoned not from rejection but from distraction, which is why abandoned-cart recovery over email and SMS often works better on mobile than on desktop.

Compare the share of mobile orders placed with an express wallet against the share of mobile sessions that reach checkout with one available. If wallets are hidden below the fold or only appear after the user enters an email, you have found the first test to run.

Speed, tested on real devices

The second structural issue is performance on mid-range phones over ordinary connections, not the fast desktop in the office. A page that loads in 1.5 seconds on your laptop can take four on the median Android, and the cost compounds across every page view in the session. We cover the mechanics in site speed and conversion rate; the point here is that you should segment performance data by device class before concluding a layout problem exists. A mobile gap that closes when you remove a heavy script is not a design gap.

Size the prize before you commit

Now put a number on it. If you lift mobile conversion from 1.4% to 1.7%, that's three tenths of a point on 140,000 sessions: 420 additional orders a month. At a $68 AOV, that's $28,560 of monthly revenue, and at a 34% contribution margin about $9,700 a month in contribution.

Compare that to buying the same orders. At 1.4% conversion you need 30,000 more mobile sessions to get 420 orders. If each session costs even $1.10, that's $33,000 a month of media for revenue that the conversion lift delivers without a variable cost. This is the actual argument for conversion work: the improvement compounds on every future session, while media spend is rented. Use the arithmetic to prioritize, and if the realistic lift is half of what you modeled, the case is still positive.

Keep AOV in the frame. Fixes that drive conversion by adding a discount or hiding shipping costs can lift the rate while hurting the order value and margin that revenue per session is built from. Judge each test on revenue per session and contribution, not conversion alone.

A sequence that works

Start by segmenting funnel metrics by device and computing revenue per session on each. Estimate the cross-device share so you know the realistic ceiling. Decompose the gap by step and point the first test at the largest one, which is usually checkout. Then work down the list: express wallets surfaced early, fewer fields, shipping cost visible before the cart, form fixes found through session replays, and performance work on mid-range devices. Re-measure revenue per session after each change, by device, for at least two full weeks.

Find your own gap

If your mobile and desktop numbers have never been split by funnel step, the biggest opportunity is probably sitting in one place you aren't looking at.