When a store tells us mobile is broken, the checkout is usually the last place we look, because on most phones the order is lost long before anyone reaches it. A checkout with a wide gap against desktop is a real problem and it is worth fixing. It is just rarely the biggest one.
The reason it gets blamed is that the checkout is a distinct thing with a beginning and an end, so it is easy to talk about rebuilding. The product page is diffuse, and the phone experience of a product page is genuinely hard to picture from a desk.
So we start by finding where the two devices diverge. Take the funnel, session to product view to add-to-cart to checkout started to order placed, and calculate each step separately for mobile and desktop. The step with the widest gap between them is the actual problem, and in most stores we look at, it is add-to-cart. People on phones are not failing to pay. They are failing to be convinced.
The phone loses the order on the product page
A phone screen shows roughly a postcard of your page at a time. Everything below that is a decision the visitor has to choose to make, and a large share of them do not.
So what is in the first screen decides the outcome. On desktop, the image, the price, the variants, the delivery information and the trust cues can all be visible at once. On a phone, the same layout usually delivers one large image and a name, with everything that answers a question far below.
That is why mobile add-to-cart rates lag. The information exists, the visitor never reaches it, and they leave believing the page did not say.
The fix is compression rather than redesign. Get price, key variant choice, delivery cost and timing, and the single most important reassurance into the first screen and a half. Move the long description below. Shorten the gallery so the buy area is not pushed off the bottom. Make the specification a short list of what buyers ask, not a supplier dump.
Measure it on the pages you changed, mobile only, and the effect is usually visible within a month.
People on phones are not failing to pay. They are failing to be convinced.
Typing is the tax
Inside the checkout, the cost is measured in taps and keystrokes. Every field is a small chance of losing someone, and phones make each field more expensive than it is on a keyboard.
So the first pass is subtraction. Company name, second address line, phone number where you do not genuinely need it, a field asking how they heard about you. Each one that is not required by the carrier or the payment processor is costing you orders in exchange for data you rarely use.
The second pass is making the remaining fields cheap. Address autocomplete removes most of the typing in one move. Correct input types bring up the numeric keypad for a postcode and the email keyboard for an email. Autofill works when fields are labelled the way browsers expect, and breaks when a theme uses custom field names.
The third is the account question. A forced account before purchase is one of the most reliable ways to lose a first-time mobile buyer. Guest checkout first, with the offer to save details after the order is placed, gets you both the sale and, often, the account.
None of this is glamorous and all of it compounds.
Wallets do more on a phone than anywhere else
Express payment methods are worth more on mobile than on desktop because they replace the part of the process phones are worst at. The address, the card number and the confirmation all come from somewhere the customer already trusts, and the sequence collapses to a couple of taps.
The placement decision matters. A wallet button in the cart or on the product page catches the shopper who is ready and does not want to see a form. Keeping the ordinary path clearly available matters just as much, because a shopper who wants to check the delivery date first should not feel pushed.
There is a reporting consequence to expect. Wallet orders often carry less customer detail into your email list, and they can bypass a discount field, so plan for both rather than being surprised.
One more thing we check. Wallet buttons that appear late, after the rest of the page has drawn, cause a layout shift that moves the buttons under the reader's thumb at the moment they tap. That single detail costs more orders than most theme decisions, and it is invisible on a fast connection.
Speed is a checkout problem in the middle of the funnel
A phone on a mobile connection is not a laptop on office wifi, and the store you test on your own device is not the store your customers use.
The pages that need to be fast are the product page and the cart. That is where the traffic is and where the decision happens. A homepage that loads slowly is a smaller problem than most stores assume, because a smaller share of buyers start there.
The usual causes are the same everywhere. Too many apps injecting scripts, oversized images served at desktop dimensions, a slider nobody uses, chat and review widgets loading before the content, and fonts that block the text from appearing.
The useful test is to load your best-selling product page on a mid-range phone, on mobile data, and count how long it takes before you can read the price and tap the button. Not a lab score. The actual experience, on the actual hardware your buyers hold.
Then remove one thing at a time and measure. Speed work is almost entirely deletion, which is why it is unpopular and why it works.
Reading the mobile funnel without fooling yourself
Two habits keep this analysis honest.
Split by traffic source as well as by device. Mobile is where social traffic lands, and social traffic browses more casually than someone who searched for the product by name. A mobile conversion rate that looks poor can be a channel mix effect rather than an experience failure, and you can only see that by splitting.
And read the funnel steps rather than the headline rate. A store-wide mobile figure tells you there is a problem somewhere. The step comparison tells you which room to go into.
One last check we always run before any of the above. Walk the store on a phone as a first-time buyer, from an ad click through to the confirmation page, and pay with a real card. Not the preview, not the theme editor. The live store, with the apps running.
We have yet to do this for a store and find nothing. Usually there are three things, one of them embarrassing, and fixing them takes less time than the meeting about whether to redesign.