A Conversation With Marina Shulga, Product Leader In Payments And Digital Infrastructure: Why Payment Infrastructure Decides Where Products Can Grow

Payments tend to be filed under operations, near the end of the process, after a customer has chosen to buy. Once a product sells across borders, that layer starts doing something larger. It decides whether a customer in another country can pay you at all, and with that, whether the market is open to you in any practical sense.

I came to this through products distributed globally, where the payment infrastructure enabled us to reach that market in the first place. You could localise the product and run the marketing, but if a customer couldn’t pay the way they normally do, none of the rest would count. The infrastructure beneath the checkout was carrying the product to market, as distribution does.

This grows harder to ignore as cross-border commerce expands, roughly three times faster than online retail as a whole. A company moving into new regions is taking on more of the exact conditions in which a home-market payment setup begins to fail.

Why Does The Payment Function So Often Stay Seen As A Cost Rather Than A Contributor?

Part of it is where the first decision gets made. When a company is formed, the choice of payment provider usually falls to finance or legal, and it depends on where the business is registered. A provider such as Stripe gives you the methods available in that home jurisdiction, which often has little to do with how people pay in the markets the company later wants to enter.

The choice is reasonable on day one, but becomes a constraint by the time expansion begins.

The other part is internal language. Payments use a great deal of specialist terminology, and the people who run them tend to report progress in those terms, which falls flat outside the team.

In recent polling of payments executives, around half still said their function is seen internally as a cost center, and a large majority said explaining payments to colleagues is a constant challenge. When the value cannot be put in terms that the rest of the business uses, it stays invisible, and so does the revenue being lost.

I place payments close to the product for that reason. Finance can tell you what the payment infrastructure costs, along with FX and if you use crypto methods, mining fees. Finance does not track the dynamics behind declines, though, or the detailed analytics on why payments fail. It does not own the technical infrastructure, the UX or the UI nor does it build the retry flows that bring a customer back for a second or third attempt after a payment fails.

The product team sits closer to all of that. Within the GDPR, it can gather data about the customer, their device, where they are and how they intend to pay, which determines whether the company can grow where it wants to.

What Does A Weak Payment Setup Block When A Platform Enters A New Market?

Three problems tend to stack up. The first is payment methods, which are deeply local. Customers in Saudi Arabia reach for MADA, customers in Brazil use PIX, customers in the Philippines use GCash, and when their preferred method is unavailable, a large share leave without buying, even when they wanted the product.

Offering the methods people in a market trust has a measurable effect on revenue. Bain has estimated that getting payment methods right can lift average order value by close to half.

The second is price. A single worldwide price ignores differences in purchasing power, so a subscription that sells comfortably in Western Europe can sit beyond reach in Central Asia or Southeast Asia, where the same figure means something quite different to the person paying.

The third is the movement of money underneath, the currency conversion and routing that carry their own costs and failure rates. None of this shows up on a headquarters dashboard. It surfaces when a company reads its own transaction data country by country, and the stakes are high, because roughly four in five shoppers say one disappointing experience is enough to send them to a competitor.

What Changes When A Company Starts Treating Payments As A Profit Centre?

The data collection changes first, and the decisions follow. While payments sit under cost, the conversation about them rarely moves past shrinking fees. Once they are read in terms of revenue, it widens to cover money lost on failed transactions that can be recovered, conversions that can be improved, and customers who stay because paying is easy.

A good deal of the work is translation. A line in a report about improved routing logic carries little weight with leadership. The same change, described as a higher share of successful payments, a lower cost per transaction, and recovered revenue that used to disappear on failed payments, gives them something to act on. The mechanics themselves carry money.

Optimising how transactions are routed can cut the cost of some card payments by as much as a quarter.

I would start with the cost side rather than with ambitious projects, because it earns credibility. The first numbers worth looking at are the decline rate, the approval rate, and the recovery rate, each split between new and returning customers. FX, chargebacks, and refunds sit outside this first look.

Read this way, the data points to revenue that can be recovered rather than fees to be trimmed, and that is what makes the case for finance. Showing them a block of recovered revenue makes the argument for the larger work far easier, and once that case lands, payments earn a place in the decisions about which markets to enter next.

The companies furthest along build that into their structure, placing payments under a cross-functional group that the chief financial and chief product officers share, so it no longer sits within a single operational team.

Where Should A Team Start If They See This Problem In Their Own Numbers?

There are three steps, and they work best in order.

Start by identifying where transactions are failing, country by country, and the reasons for the decline. Any competent payment provider has analysts who will prepare this on request. The data usually exists already and has simply never been read by the region. Once it is, the decline rates show where the setup is holding the business back.

Next, prioritise by the revenue being lost, not by the size of the market. A smaller market with a high rate of failed payments can be losing more recoverable income than a large one where almost everything goes through. The money to be won back often sits quietly in the markets, failing at checkout.

Then work through each group of declines individually, because the fix varies from one market to the next. Two cases show how differently it can go:

• Large, costly market — the issue is trust. The company already spends heavily to acquire customers, so a failed payment means it paid to bring someone in and lost the sale at the last step. The fix is usually in the data sent with each transaction: the more an issuer receives about the customer, the operation, and the service, the more it trusts a cross-border payment. That alone can add one to three points to the approval rate, and it recovers revenue without touching the marketing budget
• Smaller market — the issue is the method. Approval can fall low enough that most interested customers never pay, because they do not use cards. They pay through local alternatives. Offering the method people actually use lifts approval. The volumes remain smaller than in a market the size of the United States, but there is still revenue to be earned

The thread running through all three steps is that this is revenue the company has already paid to reach. The customers chose the product and arrived at checkout; only the payment step lost them. Reading the decline data and fixing the largest reasons before the next acquisition campaign changes what the same budget returns.