An international checkout has a particularity that single-market advice ignores: your buyers do not share one payment culture. An American shopper expects to tap Apple Pay and to see taxes appear late, a German expects an invoice, a Dutch buyer is redirected to their bank. What they do share is how carts are lost: to friction and surprise, in that order. A form that asks for information the browser already knows, a shipping cost that appears at the last step, a page that takes three seconds to show the price, and a payment that fails silently after the customer did everything right.
This page is not a list of growth hacks. It covers where money actually leaks in an international checkout, what differs on mobile, and which fixes produce a measurable change rather than a plausible story.
Express checkout: what Apple Pay and Google Pay actually change
The point of an express button is not the brand on it. It is that the wallet already holds the address, the card and the authentication, so a purchase that took two minutes of typing takes fifteen seconds of confirmation. On mobile, where every field costs a visible fraction of your visitors, this is the single largest structural improvement available to most stores.
Three implementation details decide whether you capture that value. Show the express buttons at the top of the payment step, and show them on the cart as well: their job is to let a customer who has already decided skip the form, not to decorate the bottom of it. Do not hide them behind an account wall, because the customer who would use them is precisely the one who will not create an account. And keep the card form visible below, since a meaningful share of buyers, corporate card holders in particular, still needs it.
One honest caveat: an express wallet returns the address the customer has on file, which is sometimes a work address or an old one. If your logistics depend on address quality, validate after the wallet hands it over rather than trusting it blindly.
Guest checkout and surprise costs: the two oldest leaks still dominate
Forced account creation remains one of the most reliable ways to lose a first-time buyer. The account is valuable to you; to the customer it is just another password, and they know it. The pattern that works is undramatic: let the purchase happen as a guest, then offer the account on the confirmation page, where a single click creates it from data you already hold. You keep most of the accounts and lose none of the sales.
Surprise costs are the second permanent leak, and they are cultural as much as technical. In the US, prices are shown before sales tax and the total only materialises at checkout; internationally, a customer may discover import duties after the purchase, in the carrier's hands. You cannot always change what applies, but you can always change when the customer learns it. State shipping costs and thresholds on the product page, estimate taxes as early as a postcode allows, and say explicitly who pays duties for cross-border orders. A checkout where the total only ever goes down from what the customer expected converts differently from one where it climbs.
Buy now, pay later: where it pays and where it eats margin
Klarna, Afterpay and Affirm sell the same mechanism with different centres of gravity: instalments that move the affordability decision from the whole basket to a monthly slice. In high-ticket categories with an emotional component (fashion, furniture, sports equipment), offering instalments above roughly a hundred euros or dollars changes the size of the basket a customer allows themselves.
The arithmetic on the other side is just as plain. BNPL merchant fees are a multiple of standard card processing rates. On small baskets the fee erodes margin with no measurable lift, and a store whose volume sits mostly under forty euros is paying for reassurance nobody asked for. Look at your basket distribution before signing, and negotiate per category if your catalogue spans both worlds.
One more caveat: BNPL shifts credit risk to the provider, but disputes and returns still land on you, and the refund flows are slower and less forgiving than card refunds. Read that part of the contract before the sales call, not after.
Core Web Vitals 2026: the real impact
Google's thresholds are public and concrete, and they are worth knowing individually, rather than as a composite score: Interaction to Next Paint under 200 milliseconds, Largest Contentful Paint under 2.5 seconds, Cumulative Layout Shift under 0.1. Those are limits, not stretch goals.
For a store, the page that matters is the product page, not the home page. It carries the third-party scripts, the heavy images and the blocks that shift after load. A price block that moves half a second late does not cost you a ranking, it costs a tap on the wrong button.
The fix with the best return is almost always the same: inventory your third-party scripts and justify each one. Nearly every store we examine carries three to five tags from a finished campaign or a departed agency. Removing them takes an afternoon and returns more than optimising the code you wrote yourself. Our free site check shows what loads on your page.
Returns: policy as a conversion lever
A generous returns policy raises conversion, and it raises the number of returns. Both effects are real, so the question is not whether to be generous but where the crossover sits for your catalogue. What tilts the balance cheaply is preventing the returns that come from disappointed expectations rather than from a generous policy: real size guides, photos in different contexts, honest colour rendering, concrete material descriptions. That work is one-off and it removes exactly the returns nobody wanted.
Whatever your policy is, state it before payment in one visible sentence. Uncertainty about returns is one of the most cited reasons for abandoning a cart, and it is the cheapest one to fix.
Trust signals: what an unknown store has to prove
An international buyer landing on a store they have never heard of runs a quick, mostly unconscious audit: a real company name and address, a contact channel that visibly answers, a returns address in a country they can actually reach, prices in their own currency, and reviews on a platform the store does not control. These signals weigh more the less known your brand is, which is exactly the situation where conversion is hardest.
What does not work: invented badges, ratings without a visible source, and seals that link to nothing. They are actively harmful: the visitor who sees through one starts doubting the rest of the page. Regulators have moved as well, and the FTC in the United States can now seek civil penalties over fake reviews. If you display a rating, make it clickable and let it point to a verifiable profile you do not control.
Measuring without third-party cookies
The attribution that steers most advertising budgets is less solid than the dashboard suggests, and the problem is rarely the model. It is that the measurement does not technically do what everyone assumes.
Before you compare attribution models: verify that your conversion event actually fires. Place an order yourself and check that the event arrives. We have found production sites where not a single conversion had ever been recorded, because a function was declared in the wrong scope and every guard around it silently did nothing. The dashboard showed zero, which is indistinguishable from having no customers.
Once that foundation holds, the rest becomes worth doing: measure server-side rather than only in the browser, and feed the real outcome back, returns deducted, so you optimise for revenue that stays instead of orders that come back. In high-return categories that is frequently the biggest win on the table.
Want to know where your checkout actually leaks? Describe your situation through our quote form and you get a written reply within one business day. Who we are and how we work: the agency page.
