One-tap checkout is already the default

If you are still treating checkout as a payments problem, you are optimising the one part of the journey that already works. Card entry is solved. Address entry is solved. Identity is mostly solved. What remains is a wrapper of taps, screens, forms and installs that sits between a person wanting your product and owning it — and that wrapper is where a decade of ecommerce conversion has quietly gone.

The payment stopped being the hard part

Digital wallets accounted for 56% of global online spending in 2025, and 33% of in-person spending, making them the leading way to pay worldwide.^[4] In Asia-Pacific it's 77% of online spend.^[4] Payment apps are forecast to carry 46% of global point-of-sale value by 2030.^[4]

Read those numbers as an engineering statement rather than a market statement. A wallet is a solved problem set: it holds a tokenised card, a verified billing address, a shipping address, an email, a phone number, and a biometric that authenticates the person holding the device. Every hard problem in taking money from a stranger — is this card real, is this person the cardholder, where does this go — has been packaged and handed to you for free.

Shopify reports that its own accelerated checkout lifts conversion by up to 50% relative to guest checkout, and that merely displaying it raises lower-funnel conversion by 5% even when buyers don't use it.^[6] Those are the vendor's own commissioned figures and should be read as such — but the direction is not controversial, and every large platform now reports something similar.

So if the payment is solved, and accelerated checkouts genuinely convert better, why is mobile conversion still roughly half of desktop?^[2]

The tax is everything around the payment

Baymard Institute maintains a running average of 50 separate studies of cart abandonment. The number is 70.22%.^[1] It has barely moved in years, through every checkout redesign the industry has shipped.

The reasons are more useful than the headline. In Baymard's own survey — which excludes people who were only browsing, so these are people who genuinely intended to buy — the answers cluster into three different problems that get carelessly lumped together:^[1]

Respondents could give more than one answer, so these don't sum to 100%. But the shape is clear. The pricing problems are real and mostly not fixable in the checkout — if your shipping is expensive, a better form won't save you. The trust problems have largely been solved by the wallets. What's left is a cluster of pure friction: you made me register, you made me fill in eleven fields, and then something broke.

That cluster is not a payments problem. It's an architecture problem. Every one of those causes is a thing the merchant chose to put between the customer and the wallet that was sitting there ready to pay.

Nobody abandons a cart because Apple Pay is difficult. They abandon because of the four screens you put in front of it.

Count the taps

The most useful exercise you can do this week costs nothing: pick your best-selling product, imagine a customer who has just seen it in the physical world, and count every discrete interaction between that moment and a completed order. Not the ones you think are necessary. All of them.

Two things fall out of the count. The first is that the wallet button is doing enormous work — it collapses five or six interactions into one. The second is subtler and more important: the wallet button can only start working once the customer is on the right product page. Note that the App Clip path is not magic — it still costs a scan, a card, an Open tap and a pay tap, and a fetch the customer waits through. Its advantage is not that it is shorter than a good mobile site; it is that it starts from a can rather than from a search box, and that nothing has to be installed. Everything before that point is untouched by any amount of checkout optimisation, and for a customer standing in front of a physical product, everything before that point is most of the journey.

Latency is a payment method

The taps aren't the only cost. Deloitte and Google studied 30 million sessions across 37 brands and measured what happened when mobile site speed improved by a tenth of a second. In retail, conversions were 8.4% higher and average order value 9.2% higher.^[3] Deeper in the funnel the effect was larger: product detail page to basket improved 9.1%.^[3]

• 0.1sMobile speed improvement studied^[3]
• +8.4%Associated retail conversion increase^[3]
• +9.2%Associated increase in average order value^[3]

Treat that as an association rather than a proof of causation — it's an observational study, and fast sites differ from slow ones in more ways than speed. But the effect size is large enough, and consistent enough across four verticals, that the sensible prior is: milliseconds are money, and the exchange rate is steeper than most teams assume.

This is where "one tap" stops being a design preference and starts being a payments argument. A tap is not free. It costs a network round trip, a render, a decision, and an opportunity to reconsider. The reason wallets convert is not that they're elegant; it's that they delete round trips.

The install is the tax nobody prices

Here is where the industry's favourite statistic needs pulling apart, because we think most people are drawing the wrong conclusion from it.

You will see, everywhere, that shopping apps convert around three times better than mobile web. The claim traces back to Criteo research from 2018, and we've deliberately kept it out of this article's source list because we can't stand behind the comparison. The problem isn't the arithmetic. It's that the people who install a retailer's app are not a random sample of that retailer's traffic. They are the ones who already buy repeatedly and already love the brand. Comparing them to the general mobile population and attributing the difference to the app is like discovering that people who own a gym membership are fitter than people who don't.

The honest version is more modest, and we should be careful not to commit the same error one paragraph after naming it. Contentsquare's benchmark, drawn from 99 billion sessions across more than 6,000 sites, finds returning visitors convert at 2.9% against 1.7% for new visitors — and that repeat visits are 52.8% of all traffic.^[2] Returning visitors are self-selected too, so that gap is not a clean causal estimate either. What it does establish is the direction: familiarity converts, and apps are one expensive way to manufacture familiarity for the minority who install. Earning the second order without an install or a discount is the same problem, solved more cheaply.

Apple's App Clips are the clearest expression of the idea: a small slice of an app that launches from a code, a tag or a link, without an App Store visit, with Apple Pay available immediately. Android's Instant Apps made the same bet. Mobile web with a wallet button gets most of the way there for a lot less work. The mechanism matters less than the principle — the customer should never pay an install tax to give you money. We wrote up the mechanics, including the constraint most people get wrong, in App Clips, explained for Shopify brands.

Where this is heading

Three things are converging, and all three point the same way.

1. Credentials are becoming portable

   Tokenised cards, passkeys and platform-held identity mean the merchant increasingly doesn't hold anything. That's good for you — less liability, less to breach — but it also means the account you're forcing people to create is buying you less and less. You're charging a friction toll for a service the platform now provides for free.

2. The entry point is moving off the browser

   Purchases are increasingly initiated from somewhere that isn't a search result: a camera, a short video, a message, a shelf. Each of those contexts has a couple of seconds of attention and no patience for a homepage. A journey that assumes the customer arrives at your domain and navigates is designing for a shrinking share of intent.

3. Something else may be doing the tapping

   As agents start completing purchases on a person's behalf, the checkouts that work will be the ones that are legible and don't require a human to solve a bespoke interface. It is worth noticing that "no account required, structured product data, standard wallet rails" is the same specification whether the buyer is a person in a hurry or software. Building for one-tap is, incidentally, building for that.

The strongest objection

Two, and both are fair.

"My problem isn't taps, it's traffic." Often true, and if it's true for you, this article is a distraction. Checkout optimisation is a multiplier, and a multiplier on a small number is a small number. If your store sees a few hundred mobile sessions a week, a 20% conversion improvement is a rounding error and you should go and get more people to the shelf first. The argument in this piece gets compelling at volume, or when you have a large physical footprint whose traffic you currently can't capture at all.

"Frictionless buying means more returns and more regret." Also fair, and worth taking seriously rather than waving away. Friction genuinely does filter out some purchases that would have been returned or disputed. Our honest answer is that the filter is indiscriminate — an eleven-field form removes the ambivalent buyer and the committed one at similar rates, and you don't get to choose which. The targeted alternatives work better: clear pricing before the wallet sheet opens, real product information at the point of scan, an honest delivery estimate, and a returns policy that doesn't punish. Note that the single biggest cause of abandonment is unexpected cost at 40%^[1] — the fix for regret is disclosure, not obstruction.

What to do about it on Monday

• Count your own taps. Best seller, cold device, from "just saw the product" to "order placed". Write the number down. Then do it for the path a customer takes from your packaging, if such a path exists at all.
• Split your funnel by device and put the two conversion rates next to each other. If mobile is roughly half of desktop, you're normal^[2] — and normal is the opportunity. We took the mobile–desktop conversion gap apart here.
• Delete account creation from checkout. It is one of the most-cited reasons people give for abandoning a checkout, at 18% of abandoners.^[1] Offer the account after the order, when you've earned it.
• Put the wallet button above the fold on mobile product pages, not only in the cart. Shopify's data suggests its presence alone is worth something.^[6]
• Show the full price, including delivery, before the wallet sheet opens. The largest abandonment cause is a surprise, and surprises are cheap to remove.
• Then ask the harder question: of everyone who encounters your product in the physical world this month — on a shelf, in a hand, at an event — what fraction has any path to buy it in under five seconds? For most brands the answer is zero, and that is a bigger number than any checkout test will ever return.

Sources

1. Baymard Institute, Cart Abandonment Rate Statistics — 70.22% average across 50 studies; reasons from Baymard's own survey, excluding the "just browsing" segment, multiple answers permitted. Primary ↩
2. Contentsquare, 2026 Digital Experience Benchmark — conversions — 99bn web and app sessions across 6,000+ sites. Primary ↩
3. Deloitte with Google, Milliseconds Make Millions — 30m sessions, 37 brands, four weeks. Observational; read effects as associations. Primary ↩
4. Worldpay / Global Payments, Global Payments Report 2026, Trend 3. Primary ↩
5. Shopify, Shop Pay: the best-converting accelerated checkout — figures from a study Shopify commissioned in April 2023. Vendor-reported ↩