What most people believe

A common assumption is that a card payment involves two parties - the customer and the merchant, with the payment processor as an intermediary. The payment processor enables the transaction and, for this service, takes a percentage of the payment and passes the rest to the merchant’s bank.

But this simplistic view is wrong in a couple of very important ways. A card transaction actually involves four distinct parties, and the payment processor does not control most components of the price. The role of a payment processor is closer to that of a collections agent who facilitates the collection of the fee on behalf of other parties.

Subscribe now

The four-parties in a card transaction

A card transaction involves four parties. First is the cardholder’s bank, known as the issuer. The issuer issues the card and extends credit or holds the account from which the funds come. The second is the merchant’s bank, also known as the acquirer. The acquirer connects to the card networks and settles the funds into the merchant’s account. The third is the card network. These are players like Visa and Mastercard, who operate the rails that connect the issuers and acquirers. The card network also sets critical rules that must be followed in executing this transaction. Lastly comes the payment processor: the entity the merchant has a relationship with. The processor handles the technical integration, reporting, and the merchant-facing experience of the transaction.

For every card transaction, a transaction request flows from the merchant to the acquirer, then from the acquirer to the issuer. The issuer approves the transaction, and all this communication happens on the network. Each party takes its fee out of this flow, and the amount paid by the customer, minus interchange, network fees, and the processor’s own margin, is settled into the merchant’s account.

Where the fee actually goes

The largest component of the fee is interchange. Interchange is a charge by the issuer, not the processor. This is the fee the acquirer pays to the issuer for payment using the issuer’s card. Card networks have a predefined rate table for it. Interchange depends on the card type, merchant category, transaction method, and region, among other factors. Interchange is typically much higher for credit cards than for debit cards, primarily to account for chargeback risk and the cost of funding the loyalty programme.

The second component is the network fee, charged by the card networks. This fee covers the use of the rails and is typically much smaller than interchange. It is also less transparent.

The third component is the acquirer margin. The processor controls this component and can customise it for each customer. Whatever room the processor has to offer a discount generally comes out of this component of the overall fee.

Subscribe now

Another layer of fee is added when the transaction is cross-border, i.e., when a card issued in one country is used with a merchant in another country. In such cases, a currency conversion fee is charged in addition to the standard fee structure. This fee is usually in the range of 1 to 2%. It covers not only currency conversion but also the higher fraud risk and the more complex settlement process.

Why this structure persists

This structure for card settlement has remained the same for decades. The reason is that no single party has both the incentive and the ability to change it unilaterally. There have been some regulations, especially when merchants have fought for them. The USA capped debit interchange for large banks in 2011. The EU capped interchange on consumer credit and debit cards in 2015. Both of these reduced interchange revenue for issuers and pushed them to look at other sources of revenue to compensate for this drop.

These interventions demonstrate something important: interchange is not a natural market price. It is a rate set by the networks because of their power. When regulators intervene, interchange moves substantially. But these interventions are specific and narrow, usually covering specific jurisdictions or card types. The networks have the greatest negotiating power in the transaction flow, and they naturally end up controlling most of the pricing in the market.

Subscribe now

What this means for a payments company

A payment processor or merchant acquirer charges its customers a price that it does not control, except for a thin margin. This has three consequences.

First, competing on price is a bad strategy for a payment processor because it does not control most of the price. When a processor advertises an aggressively low rate, it is usually operating at zero or negative margins, funding it by cross-subsidising from another product line, or recovering the cost elsewhere through monthly commitments or chargeback fees. This follows the same pattern as cross-border pricing, as covered in Memo #1. The stated fee is not the real fee, and the components hidden from the customer actually determine whether the business is sustainable.

Second, the payments company needs to build its differentiation to protect its margin, which typically sits in the range of 0.3% to 0.5%. For a payments processor, that differentiation lies in better fraud tools, faster settlement cycles (covered in Memo #7), easier integrations, or better reporting. Building an efficient and differentiated service means the company retains healthier margins on the same fee structure over the long run.

Third, processors can improve their margins by understanding how interchange is categorised. Transactions qualify for different interchange rates based on criteria such as whether the card was present, whether address verification was performed, and the merchant type. If a processor helps merchants qualify for the lowest applicable interchange rate, they reduce the interchange cost passed on to the issuer and expand their own margin without touching the fee.

Without understanding the numbers that make up the cost of card processing, you cannot compete for the 0.4% that is actually yours.

Not surprisingly, every deal announcement characterises the deal as strategic. Something that will add scale or distribution, enhance stablecoin infrastructure, or strengthen licensing footprint. While these are partially true, the larger reason is much simpler. Current valuations make these companies look cheap, and buyers with capital are coming in. Fintech M&A multiples were 7.7x revenue in 2021 and now stand at just 4.4x. Strategy is definitely part of the story, but the more compelling reason why these deals are happening now is their valuation.

Subscribe now

This is not new. This has happened whenever quality companies have become available at a discount. What happens with this is that buyers underestimate the cost of integrating the company they are buying.

Bank of America bought Countrywide in 2008 for $4 billion. This was a steep discount compared to Countrywide’s valuation a couple of years back. BofA wanted to get scale, distribution and a mortgage servicing platform, which would otherwise have taken years to build. But together with these assets, BofA also inherited the legal exposure of Countrywide’s underwriting practices. Over the next decade, BofA paid more than $40 billion in settlements, litigation and other losses. A big part of this was the price BofA paid to absorb a business that, by design, was not fit to run inside BofA.

Buying the company is the first step. Integrating it to realise the full value is the main step. This is usually a multi-year and multi-department project. And in payments, it is much harder than any other industry.

Payments acquisition is complex because it means merging two different compliance stacks built on different vendors and calibrated to different risk appetites. It also means reconciling between two networks of banking relationships which are influenced by a gamut of regulations. It also means combining licensing footprints across multiple jurisdictions, identifying synergies and redundancies in each jurisdiction, and finally merging the two product architectures that may not be compatible in their build while continuing to serve customer transactions.

The full integration usually takes two to three years. However, most models used to calculate the acquisition price bake in a much smaller timeframe. The more the gap between the modelled timeline and the actual time taken, the more value is lost from the deal. This manifests in duplicated teams that cannot be merged, engineering effort spent merging the two platforms instead of building new products, or losing customers whose experience gets worse during the transition phase.

Subscribe now

A discount in the valuation of a company could also be hiding something; usually, when a company is available at a fraction of its former valuation, it happens for some reason. This could be weakening unit economics, high churn in the customer base, a lingering compliance debt, or a product that is no longer competitive. While the acquirer prices the deal on a revenue multiple and assumes the other issues are easy to fix, these fixes almost always take longer and cost more than planned.

This is not a case against acquisition or consolidation. Many of the deals work well, especially where the acquiring company is honest about why the target is cheap and realistic about the integration effort and costs. This is more true in the current scenario: the deals most likely to succeed and increase in value will be the ones not with the best strategic narrative, but the ones rooted in a realistic understanding of the target’s valuation and integration effort.

A discount is not a bargain until you know what is leading to the discount. This is especially true for payments M&A.

This is the most important customer acquisition channel in payments, as it is the only way to acquire customers at scale. Enterprise sales are slow and require a one-at-a-time effort. Self-serve is cheaper but still acquires customers one at a time. On the other hand, a single platform partnership can bring in 10,000 or 100,000 new customers under a single agreement. For a payments company trying to reach scale quickly, partnerships look like the magic bullet.

But payments partnerships are also not simple to execute. Most companies go through several partnerships and spend significant time and money before getting them right. It is critical to get the mechanics of a partnership right for it to succeed. The strategic case for this playbook, and why it cannot be run simultaneously with the others, is covered in Memo #4. This Memo covers the operational mechanics in depth

.What most companies believe

The mental model most companies bring to a partnership is transactional. They think of the platform as a distribution channel to get more customers.

This view misses two things. First, and most importantly, the platform is not a distribution channel. It is a customer in its own right, with its own procurement processes, risk management function, compliance requirements, and commercial priorities that are rarely fully aligned with yours. Second, the integration is not just a technical exercise. It is an operational alignment on both sides, ensuring that both partners reliably meet each other’s requirements at scale.

Subscribe now

How deals get structured

Payments partnerships come in three structural forms. The choice of structure determines how the relationship should be run.

A referral arrangement is the simplest model. The platform directs users to the payments company and receives a referral fee. The payments company owns the customer relationship. A typical example is a payroll software company that refers its SME customers to a cross-border payments provider for international salary payments. While simple to implement, platforms with meaningful volume rarely accept them. They prefer an arrangement that allows them to capture more of the economic value in the partnership.

A white-label arrangement goes deeper. The payments company provides the infrastructure, and the platform brands it as its own. The platform owns the customer relationship. The customer never sees the payments company’s name or brand. Revenue is split through a combination of a flat per-transaction fee and a share of the FX margin. This model is popular among challenger banks and neobanks for their international transfer products. The customer transacts on the bank’s interface, while a third-party payments company processes the transaction behind the scenes. Airwallex has built a significant portion of its business on this model, powering the embedded finance and FX products of other fintechs.

Subscribe now

A tech-native arrangement is the deepest form of partnership. The payments company’s product is integrated into the platform’s core workflows: the payout flow, the reconciliation system, and the reporting dashboard. The customer experiences it as a native part of the platform rather than a third-party service. Payoneer’s integrations with Upwork and Fiverr are good examples of how to do this at scale. This form of partnership is the most complex to build and the hardest to exit.

The commercial structure of a partnership typically involves three negotiated elements.

Revenue share is the percentage of each transaction’s revenue that goes to the platform rather than the payments company. In a well-negotiated deal, the payments company retains 60-80%. In a poorly negotiated one, this can fall to 20-30%.

The pricing floor is the minimum rate the payments company will accept, below which the economics don’t work, regardless of volume. This is the number that must be known before the first commercial conversation, and not estimated on the fly.

Exclusivity and volume commitment are the most consequential commercial terms in any partnership agreement and the ones most often negotiated away too easily. An exclusive arrangement means the platform routes all its payouts through a single provider. This is rare and increasingly hard to achieve because platforms understand that single-provider dependency is an operational risk. But where it is achievable, typically with a new platform at lower volume or when the payments company is investing in the integration, it can be enormously valuable.

A non-exclusive arrangement means the platform can use multiple providers simultaneously. Most mature platforms operate this way. In some cases, the platform allows its customers to choose their provider. More often, the platform retains routing control. The key question, then, is how the volume is allocated among providers.

Subscribe now

Platforms typically allocate in one of three ways. It can be corridor-based, with different providers serving a defined corridor. This creates predictability of business but limits growth. Or a performance-based allocation, where the platform monitors operational metrics like settlement success rate and payout speed and shifts volume to the provider with better performance. While this rewards operational excellence, it also creates constant competitive pressure. Or a rate-based auction, which is the worst outcome for the payments company. Here, the platform asks for the best rate at that moment and routes each payment to the provider offering it. This turns every transaction into a commodity auction and completely eliminates margin stability.

In practice, a non-exclusive arrangement without a volume commitment means the payments company has no guarantee of any business. The platform can route 100% of its volume through a competitor without breaching the contract. This is why volume commitment is a non-negotiable commercial protection, and not something to give up for a faster close.

Where deals break down

A payments partnership moves through five stages before going live. Understanding at which stage a deal is lost reveals exactly what is lacking in the company’s partnership capability.

Stage 1 is commercial negotiation. The platform has talked to a few providers before you and knows what the market offers. If the revenue share you can sustain at your pricing floor does not meet the platform’s expectations, the conversation ends here. Most early-stage pipeline attrition happens at this stage. Understanding the market dynamics and your commercial floor before the first meeting is required to navigate this well.

Stage 2 is compliance alignment. The payments company’s KYC standards, transaction monitoring thresholds, and acceptable use policies need to be compatible with the platform’s customer base. The platform’s customers may have a different risk profile than the payments company’s existing portfolio. Reconciling these takes repeated back-and-forth between legal and compliance teams on both sides. Every identified gap requires a negotiated solution that must then be planned and built.

Stage 3 is technical integration. The API integration itself can be straightforward. But the work of mapping the platform’s data structure to the payments company’s settlement reporting, building the reconciliation bridge between the two systems, and managing edge cases can become real bottlenecks.

Stage 4 is the pilot period. Most partnerships start with a live pilot before full launch. A subset of the platform’s volume is routed through the payments company’s rails and monitored for operational metrics. Pilots reveal gaps that remained hidden during integration and may take a long time to fix.

Stage 5 is launch and handoff. Even after the pilot is successful, the transition to full volume can introduce new operational stress. High volumes can test the infrastructure’s stability; support queries can be much higher than the planned capacity; and new reconciliation gaps can appear after the first full cycle. Post-launch is the beginning of a new operational phase.

What makes a partnership work

The partnerships that grow in volume and prove sustainable have three structural characteristics that have nothing to do with the product.

The first is a shared definition of success at 90 days, six months, and twelve months. Partnerships without an agreed scorecard create misaligned expectations between the partners. The platform may measure payout success rates, while the payments company measures volume and revenue share, and neither knows whether the other is satisfied. This is basic but consistently overlooked.

The second is arriving at commercial terms that create genuinely aligned incentives on both sides. Sometimes this can mean leaving some money on the table. A baseline for aligning incentives is a revenue-sharing model in which the payments company earns nothing on failed transactions. Another version of this includes penalties if settlement success rates fall below a threshold. Volume commitments that, when met, trigger investments in payment infrastructure create reciprocal obligations. When commercial terms are structured this way, both partners have an incentive to fix problems rather than find who to blame.

The third is a named internal champion on each side with both the authority to make decisions and the accountability for the outcome. There should be one person on the platform who needs this partnership to succeed, and one person at the payments company who owns it end-to-end. Without this, partnership efforts risk losing priority and momentum.

What this means

Payments partnerships are the highest-potential and most poorly executed GTM channel in the industry.

This has three consequences.

First, pipeline size is the wrong measure of partnership capability. A company with 40 active conversations and 2 live integrations is doing worse than one with 10 conversations and 5 live integrations. The right diagnosis is understanding the stage distribution at which point deals are consistently dying. Dying at commercial negotiation means the pricing floor is not understood or not competitive. Dying at compliance alignment means the compliance framework was not built for platforms. Dying at technical integration means the product was not designed for third-party distribution. Each is a structural problem that needs a structural fix, without hiring a larger BD team.

Second, the post-launch phase requires as much investment as the pre-launch phase. Most partnership teams are built to close deals. They are rarely resourced to actively manage the relationship as it matures. Sustained investment in post-launch partnership management is a must to keep partnership volume growing.

Third, every non-exclusive arrangement with a growing platform will eventually face competition from within the account. The platform will certainly sign a second provider. The only question that remains is whether the payments company continues as the primary provider or is replaced by a competitor.

The partnership playbook is the fastest way for a payments company to scale. It is also the easiest way to stay busy without any real growth. Knowing which one you are doing requires measuring the right things at the right stages.

Three reasons why subscription pricing in payments is less clever than it looks.

The first is cost structure. In SaaS, the marginal cost of serving one more customer is close to zero. The software is already there, and adding another user costs almost nothing. Subscription pricing fits because the cost structure is flat. If you are making a product like Salesforce, your costs change very little whether a user logs in for 5 hours or 150 hours per month. In payments, on the other hand, every transaction has a variable cost associated with it. Subscription revenue is fixed, but costs are not. This imbalance can work in some cases, but not where margins are already thin.

The second is the portfolio cross-subsidy. A customer sending $10,000 a month on a $99 subscription and a customer sending $100,000 a month on the same plan are not the same. One pays an effective rate of 0.99%, while the other pays 0.10%. The low-volume customer is subsidising the high-volume one. This model holds when the portfolio skews towards smaller customers. As the business matures and wins larger customers, the subsidy runs in the wrong direction, and the unit economics deteriorate.

Subscribe now

The third factor is stickiness. In SaaS, subscription pricing reinforces lock-in over time. The longer a customer stays, the more data and workflows accumulate within the product. This makes it more expensive for a long-standing customer to switch than for a new one. And this makes subscriptions attractive. The subscription pricing and product stickiness reinforce each other, creating a flywheel. In payments, this lock-in effect does not exist. Stickiness in payments comes from factors that are unaffected by the pricing model. A customer who has been on a payments subscription for three years is no harder to lose than one who signed up last quarter.

What does this mean for a payments business?

Revolut Business and Wise have taken opposing positions on this issue. Revolut offers four subscription tiers, each with different limits on free FX transfers. Wise uses a purely variable model: you pay per transfer and nothing else. Both have been successful, but for very different reasons.

Subscribe now

But Revolut does not sell just a payments product. It sells an entire platform: cards, expense management, approval workflows, accounting integrations and analytics. Many of Revolut’s subscribers use it primarily for spend controls and expense tooling, not for FX transfers. Their international payment usage is occasional. The subscription economics work because the cost of heavy FX users is offset by subscribers who rarely use the transfer feature.

Wise’s portfolio has no such cross-subsidy. Every customer is there specifically for FX usage. They will all maximise usage and optimise their per-transaction cost. The FX product does not give Wise the customer mix that makes subscription economics viable. This is why Wise charges per transfer and has not tried a flat subscription.

The implication for a pure-play cross-border payments company is straightforward. Replicating Revolut’s subscription model without Revolut’s product breadth means attracting all the heavy FX users and none of the casual subscribers to offset them. The economics will fail.

The only scenario where subscription pricing makes sense in payments is when pricing is mainly for platform capabilities and not transaction volume. A multi-currency treasury dashboard, an approval workflow engine, and a compliance reporting tool are all examples. Here, the subscription provides access to fixed-cost products and not to variable-cost processing. That makes it a viable model. But bundling transaction volume in a subscription and hoping the mix works out does not.

Subscription pricing looks glamorous and modern. But it is a margin problem disguised in a fancy dress.

This is the second of two Memos on compliance in payments. Memo #6 covered how the compliance stack is built in fintechs, outlining the five layers from KYC through regulatory reporting, and why fintechs often underinvest in each of them. This Memo covers where to look for the cost of inadequate compliance, why it is easy to misattribute, and what it would take to measure it honestly.

What most people believe

Most fintechs measure their compliance function’s performance by regulatory outcomes. If there are no fines and no regulatory notices, compliance is working well.

This is a dangerously incomplete picture. Regulatory outcomes measure one dimension of compliance: whether the company meets its minimum legal obligations. They say nothing about whether the compliance function is supporting or undermining the business. It’s possible that even a fully compliant company can churn customers, lose banking partners, and shrink its revenue growth because its compliance function is underperforming.

The real cost of inadequate compliance goes well beyond the fines and notices received. The real cost is a slowdown in business growth because compliance is too slow, too manual, or too opaque to support it.

Where the costs surface

The customer experience cost

The most direct cost of inadequate compliance is damage to the customer experience at the two most critical moments: onboarding and transaction processing.

For onboarding, the typical failure occurs when the KYB process is not designed from the customer’s perspective. A badly designed process looks like this: the application flow requests documents the customer was not told to prepare, the compliance team requests additional documents, and the customer resubmits. The compliance team again requests clarifications that weren’t clear in the ask. Each round takes two to three business days because the review queue is manually managed. After a few rounds, the customer has lost all inclination to work with the fintech.

This gets recorded as an incomplete application. But it is actually a customer lost due to a poorly designed onboarding flow, one built around the compliance team’s process rather than the customer’s journey.

In transaction processing, the typical failure occurs when a transaction goes into a compliance hold. A payment is flagged by the transaction monitoring system and placed on review. The customer receives a templated notification that their transaction is under review and that they will be contacted if further information is needed: no timeline, no explanation and no personalisation. The manual review takes three to five days to clear. The customer calls customer support, who cannot access the case details because the compliance and support systems are not integrated. The customer escalates. But by the time the hold is released, the lack of funds has resulted in a missed financial commitment. The relationship is now damaged.

Subscribe now

The banking partner cost

Banking partners, by design, have an interest in the quality of the fintech’s compliance programme. They usually conduct periodic reviews to assess it and ensure there are no breaches.

If a banking partner observes gaps in compliance processes, like high false-positive rates, SAR filing backlogs, or gaps in regulatory reporting, these will come up during their review. The partner’s response options range from requiring a remediation plan to adding restrictions on corridors or customer types to terminating the relationship. The last option, though rare, can be existential for the fintech.

The more common outcome is that the banking partner imposes certain requirements on the fintech. These can be in the form of restrictions on specific customer categories, caps on transaction volumes in certain corridors, or more frequent reporting obligations. These requirements can directly constrain the business by limiting the corridors, customers, and transaction types the fintech can serve. The revenue loss because of these restrictions is rarely attributed to inadequate compliance. It is generally shown as a commercial headwind with no root cause or solution.

The revenue cost

Another cost of inadequate compliance is that it shrinks revenue in ways that are genuinely hard to measure. The difficulty of measurement means that this aspect of the cost is consistently overlooked.

The most significant revenue shrinkage is due to lower onboarding conversion. A business customer who abandons the KYB process midway is not a lost compliance case; they are a lost revenue opportunity. If the average revenue per business customer is $3,000 annually and the fintech’s onboarding completion rate is 50% against a competitive rate of 70%, the gap represents foregone revenue due to a poor onboarding experience.

The second shrinkage occurs in the volume from existing customers. A business that has experienced a compliance hold often reduces its transaction volume as a precautionary measure. Unexpected delays in the movement of funds are costly for any business managing cash flow. Rather than routing all of its cross-border payments through the fintech and risking cash-flow disruption, it splits volume across two providers. The customer’s share of wallet drops, but this drop is rarely connected back to the compliance experience that caused it.

The third suppression is in customer referrals. B2B payments are a relationship-driven market. Businesses recommend providers to their networks when they are satisfied with them. A business that had a poor compliance experience does not.

The regulatory cost

The regulatory cost is the one that gets measured, and often the only one, because it is impossible to ignore. A regulatory action, a fine, or a licence restriction has a number that is straightforward to calculate.

But even here, most companies underestimate the full cost. The direct penalty is the most visible figure, but it represents only part of the total. The remediation programme required after any regulatory action typically costs significantly more: external auditors, compliance consultants, technology upgrades, and additional headcount, all under stretched timelines that drive costs higher. These are rarely attributed to inadequate compliance and are treated as one-time project costs.

The reputational cost is harder to measure but equally important. A regulatory action is public. All future conversations with banking partners, investors, and enterprise customers will involve managing the fallout.

Subscribe now

Why the cost gets misattributed

The primary reason the cost of inadequate compliance is consistently underestimated is that it is distributed across the organisation rather than aggregated in one place.

Churn from compliance friction shows up in the customer success team’s retention reports and is misattributed to product dissatisfaction or competitive pricing. Onboarding drop-off shows up in the acquisition team’s funnel analysis, attributed to friction in the application flow. Banking partner restrictions appear in the operations team’s corridor reports and are attributed to banking relationship issues. The compliance team, whose metrics focus solely on regulatory outcomes, has no visibility into the business impact of its decisions.

This is why compliance should be treated as a core operational function. Inadequate compliance affects GTM, operations, and product - all at the same time, without any single team seeing the full picture.

What honest measurement looks like

To measure the impact of inadequate compliance honestly, companies need to track compliance as a business function, not just a legal one.

• They should track onboarding completion rates by customer type and by compliance decision point to see exactly where in the KYB process customers are abandoning.

• They should track compliance hold rates, hold duration, and post-hold churn to see the direct relationship between holds and customer churn.

• They should track false-positive rates in transaction monitoring and the cost of manual review to quantify the benefit of investing in better rule calibration.

• They should track changes in customer volume in the 90 days following a compliance interaction to see the indirect revenue impact.

These are business metrics that directly measure the output of the compliance function. Tracking them requires the compliance, product, and commercial teams to share data and accountability in ways that most fintechs are not yet structured to support.

What this means

Inadequate compliance has three costs that compound over time and reinforce each other.

The customer experience cost is immediate. Customers who experience poor compliance interactions churn, refer less, and reduce volume. This cost is recoverable if the underlying infrastructure is improved. But delays lead to this cost compounding heavily.

The banking partner cost is slower to materialise but much harder to reverse. A banking relationship flagged for compliance concerns takes years to recover fully. A banking relationship that ends cannot always be replaced in the same corridors at the same cost.

The regulatory cost is the most visible. The remediation and opportunity costs that follow a regulatory action typically dwarf the penalty itself.

All three costs have one common characteristic: they are preventable. They can be prevented by investing in the right compliance infrastructure, as covered in Memo #6. Companies that build the compliance stack properly avoid regulatory problems while building a compliance function that actively supports growth rather than constraining it.

The result of poorly done compliance shows up everywhere: in customer churn data, in long support queues, in tough conversations with banking partners, and in slowing revenue numbers. You simply have to know where to look.

At the time, it sounded like a defensive answer from someone who did not want to acknowledge the threat. A decade and a half later, I realise he was more right than I gave him credit for.

People trust financial companies with the one thing they cannot afford to lose. Their money – in the form of their savings, investments and payments. When a fintech does any of these, it is doing much more than selling a service. It is asking for trust. And a licence is the most credible signal of that trust.

Trust is the hardest asset to replicate. A well-funded competitor can replicate everything else in 12 months: a better UI, a faster onboarding flow, a cleaner dashboard, and a lower fee. But a regulatory licence, built across multiple jurisdictions and supported by the right compliance infrastructure, banking relationships, and regulatory track record, takes many years to build. This is the hardest barrier to entry for new competition, and doing it well is the most durable competitive moat.

Subscribe now

But let’s not mistake compliance for the paperwork that supports the licence. Compliance is the operating layer that makes the licence effective. The KYC process to verify the customer, transaction monitoring to ensure the platform is not used for fraud or money laundering, sanctions screening to keep the company and its customers away from restricted counterparties, and regulatory reporting that demonstrates credibility to the regulator, year after year. All of this makes trust possible. The product and the interface on top of it, however well designed, are the wrapper.

The best evidence of this is what happens when the world’s best technology companies try to enter financial services. Apple Pay, Google Pay, and X’s payments ambitions all share a common characteristic: they are built on top of licensed financial institutions. Apple and Google work through card networks and issuing banks. X’s lofty financial services ambitions have started with a partner bank that holds customer deposits. These are not small companies with limited resources or engineering talent. They are the most capable technology organisations in the world. And yet, when it comes to financial services, they build within the regulatory infrastructure, not outside it.

Subscribe now

Some see this as a system protecting incumbents. That is a superficial and incorrect reading. The licence requirement is not a moat that the Citibanks of the world built to slow down competition. It is the safeguard regulators built to protect the people whose money is at stake - the customer depositing their savings, the small business sending a payment to a supplier, and the worker sending money home to their family. These are people who cannot afford to have the system fail.

The fintechs that understand this treat compliance not as a function that permits them to operate, but as a signal they send to customers every day. That signal is: your money is safe here. We have the licence, we have the track record, and we have a compliant infrastructure. Everything we have built, we have built on top of that.

The Citibank executive was not being defensive. He was being precise. The fintechs that eventually threatened him did not win by replacing the licence. They won by earning one.

How you build compliance in the early years determines what you can scale later. The right way is to build compliance into the product from the start. Otherwise, it will need to be rebuilt again as soon as you hit scale.

What most people believe

The common assumption is that compliance is a people problem. Hire a good Chief Compliance Officer, get a team underneath them, and the function will run smoothly. The CCO knows the regulations, and the team does the processing.

This is what compliance looks like from the outside. From the inside, it’s very different.

Compliance in a fintech is a combination of people, technology, and processes, each of which has to be deliberately designed and built. The CCO cannot do much without a KYC system for new accounts. The KYC system cannot do much without a robust transaction monitoring system. The transaction monitoring system cannot do much without a sanctions screening engine to identify restricted counterparties in real time. These are part of a single stack, with each layer depending on the one below it.

The Compliance Stack

Layer 1: KYC and KYB

Know Your Customer (KYC) and Know Your Business (KYB) are the starting points of the compliance stack. Before any customer can transact with a fintech, it needs to verify their identity.

KYC means ID verification, liveness checks, and a database match against government watchlists. Solutions for these are now available off the shelf. Vendors like Jumio, Onfido, and Persona provide API access for document scanning and facial recognition. A well-built consumer KYC flow takes under five minutes and requires no human review for most applicants.

However, for KYB, the solution is not seamless. KYB requires collecting and verifying entity formation documents, understanding the ownership structure, identifying the Ultimate Beneficial Owners who own more than 25% of the entity, and running each UBO through the same checks as an individual consumer. For a simple company with two shareholders, this is manageable. For a holding company with subsidiaries across three jurisdictions, it can take weeks.

The vendor landscape for KYB is also less mature than for KYC. Providers like Middesk, Moody’s and Kompany offer business verification, but the data quality varies significantly by country. Complex structures require extensive manual review to verify submitted documents.

The build-vs-buy decision for this layer is straightforward: buy the solution and build the workflow logic. The rules of when to auto-approve, when to flag for manual review, and when to auto-reject are specific to the company’s risk appetite and regulatory obligations.

Subscribe now

Layer 2: Sanctions screening

Sanctions screening checks every customer and counterparty against lists of restricted individuals, entities and jurisdictions. The primary repositories of sanctions lists are OFAC and the UN Security Council. A fintech cannot transact with anyone covered in these lists, and a failure to comply attracts heavy penalties.

Sanctions screening solutions are quite mature. Providers such as Dow Jones, LexisNexis, and ComplyAdvantage maintain up-to-date watchlists and offer API access for real-time screening. When a transaction is submitted for review, it is screened, and results are returned in milliseconds.

The operational challenge here lies in managing the false positives. Watchlist matching is imprecise by design because the lists contain names that are common. The compliance team has to review each flag and determine whether it is a true match or a false positive. At low volumes, this is manageable. At scale, a false positive rate of even 1% means thousands of manual reviews per day. Building a workflow that triages these flags efficiently and routes them to the right reviewer is as important as the screening technology itself.

Layer 3: Transaction monitoring

A transaction monitoring system reviews every transaction after onboarding for patterns that might indicate money laundering, fraud, or terrorist financing.

The simplest version of implementing transaction monitoring is rule-based, i.e. any transaction above a certain threshold, any customer who sends more than a certain amount in a given period, or any payment to a high-risk jurisdiction is flagged. These rules are quick to build but also easy to game. They generate significantly higher false positives.

The more sophisticated version uses machine learning models trained on historical transaction data to identify behaviour that is anomalous relative to a customer’s established pattern. A customer who typically sends $2,000 per month suddenly sending $50,000 to a new counterparty in a new country is an anomaly. A rule-based system may or may not catch this depending on where the threshold is set. A behavioural model will flag it because it deviates from the customer’s baseline.

Most early-stage fintechs start with rule-based monitoring because it is faster to build. The problem is that rule-based systems produce alert volumes the compliance team cannot keep up with, and they miss sophisticated patterns operating just below the rule thresholds. Moving to a more sophisticated model requires clean transaction data, which takes time to accumulate, and data science capability, which most compliance teams do not hire for.

The build-vs-buy decision here is more complex than at the KYC layer. Vendors like Featurespace, NICE Actimize, and Sardine offer transaction monitoring platforms. But configuring these for a specific business model and customer population requires significant work. A remittance company and a B2B payments platform have very different risk profiles and require vastly different configurations, even when using the same vendor.

Subscribe now

Layer 4: Suspicious Activity Reporting

When the transaction monitoring flags a transaction, and the compliance team cannot rule it out as a false positive, the company has an obligation to file a SAR (Suspicious Activity Report). In the US, it is filed with the FinCEN; in the UK, with the National Crime Agency; and in India, with FIU-IND.

SAR filing requires a judgement call on whether there is reasonable suspicion of financial crime. Filing too broadly wastes regulatory resources, and filing too conservatively is a regulatory violation. The threshold is deliberately kept vague, implying that the decision depends on the CCO’s experience and judgement.

The process around SAR filing has to be airtight. The investigation must be documented, and the decision to file or not must be clearly recorded. And critically, the customer cannot be informed that a SAR has been filed against them. Tipping off a customer is a criminal offence in most jurisdictions.

This process is complex. Managing it with a spreadsheet and a shared inbox fails at any meaningful scale. A proper case management system that can track investigations, store evidence, record decisions, and manage filing deadlines is the minimum requirement.

Layer 5: Regulatory reporting

Beyond SARs, payments companies have ongoing regulatory reporting obligations. Some of them are Currency Transaction Reports for cash transactions above certain thresholds, cross-border transaction reports, periodic attestations about the effectiveness of the compliance programme, and responses to information requests.

This layer is the least glamorous and the most frequently neglected. It is also the layer where regulators most easily find deficiencies. A company with sophisticated monitoring but poor regulatory reporting will still fail an examination.

Why do most fintechs underinvest

The compliance stack described above takes time and money to build. At the seed and early Series A stage, most fintechs build the minimum required to obtain a licence and launch: a basic KYC flow, a simple sanctions screening integration, and some manually reviewed transaction monitoring rules. This makes sense when the priority is time-to-market.

The problem is that “minimum to launch” becomes “permanent state” at too many companies. While headcount grows, the compliance team’s tooling does not keep pace. Manual processes set up at a scale of 1,000 customers become bottlenecks at 50,000 customers.

The trigger for investment is usually external, when a regulator flags deficiencies or a banking partner demands evidence of a robust compliance programme before renewing the relationship. These external triggers then force the investment that should have been made earlier, often at higher cost and under time pressure.

What most companies don’t realise is that underinvestment in the compliance stack surfaces directly in customer experience, banking partner relationships, and eventually, revenue. Memo #7 will cover the full cost of inadequate compliance.

What this means

Compliance, as a function, is widely misunderstood in fintech. It is not a function to translate the regulations and tell what can and cannot be done. It is a critical operational muscle the company needs to serve customers, move money, and partner with banks.

This has three consequences.

First, the GTM playbook determines the compliance architecture, as covered in Memo #4. A self-serve platform needs automated KYC to process thousands of applications without human review. An enterprise sales model needs bespoke onboarding flows and dedicated compliance handling for each client. A marketplace partnership model needs compliance frameworks that operate at the platform level rather than at the individual customer level. The wrong compliance architecture for your GTM model is as expensive as building the wrong product.

Second, compliance friction is a direct cause of churn, as covered in Memo #2. Customers who are placed on hold with no communication, or who wait weeks for a review decision, will leave. The quality of the compliance experience is a product decision, and it should be measured as such.

Third, the cost of fixing the underinvestment continues to grow with scale. A compliance stack that is adequate for 10,000 customers will break at 100,000 customers. Rebuilding it with the regulator or a banking partner closely watching over it, under a strict deadline, and keeping the product live is the most expensive way to rebuild it.

Most compliance problems are product and infrastructure problems, and compliance takes the blame. The right stack is required for the function to do what it is supposed to – let the business grow without breaking.

It did not turn out that way. Bitcoin is now digital gold - a store of value, rather than a medium of exchange. The volatility that makes it attractive as a speculative asset makes it useless for payments. You cannot price a cross-border invoice in something that can move 15% in a day.

Stablecoins now claim to finish what Satoshi started. They are price-stable, blockchain-native, and fast. Their promise is compelling enough that serious people have begun arguing that SWIFT’s days are numbered. Yet stablecoins have inherited Bitcoin’s original problem in a different form - hype.

The hype around stablecoins is based on a misdiagnosis. The misdiagnosis is that the bottleneck in cross-border payments lies in the messaging layer. SWIFT is a messaging network, and yes, a very old one. But upgrading the messaging layer solves roughly 10% of the problems. The other 90% is everything else: compliance screening, FX conversion into local currency, last-mile delivery to the recipient’s account, and regulatory reporting in both jurisdictions. All of these remain even after you change the rail.

Let’s say someone sends a $50,000 payment from the US to Vietnam. Stablecoins can move USDC between two wallets in seconds. But the recipient in Vietnam does not want USDC. They want VND in their bank account, with legally compliant documentation. This requires a licensed entity in Vietnam with banking relationships, local FX liquidity, and the compliance infrastructure to process the transaction legally. All of this is identical to what traditional cross-border payments require. While stablecoins skip the middle correspondent banking chain, they do not eliminate the hard parts at either end.

Subscribe now

Compounding this is a liquidity problem the hype consistently ignores. Stablecoins need deep liquidity across every corridor they serve. This liquidity exists for major currencies but not for emerging ones. Who provides deep USDC-to-NGN liquidity at 2 AM in Lagos? Either a centralised market maker who will charge for the service, or nobody. Thin liquidity erodes the cost advantage that stablecoins promise. And unlike SWIFT, where a failed transaction has established legal frameworks and dispute resolution mechanisms behind it, a stablecoin transaction that gets stuck leaves you with no counterparty to call. The absence of intermediaries is a defining feature of crypto. In cross-border payments, it increases the risk manifold.

As regulation catches up, compliance requirements will increasingly mirror those of established payments companies. The cost advantage will narrow. A licensed, compliant stablecoin provider will need much the same infrastructure as everyone else.

This does not mean stablecoins have no role. There are two use cases where the case for stablecoins is genuinely strong. The first is wholesale settlement between institutions that both natively hold stablecoins. Crypto exchanges settling with each other is the clearest example. They have no conversion at either end and no last-mile problem. The second is in corridors where traditional infrastructure is broken or absent. In parts of Africa, Southeast Asia, and Latin America, a stablecoin rail with a functioning local off-ramp can outperform the alternatives. Stablecoins help leapfrog the existing infrastructure with something much better.

Outside these two cases, stablecoins will compete on the same dimensions as everyone else – cost, speed and reliability. This is the honest version of the stablecoin payments thesis. It won’t become a SWIFT killer, but a specialised rail for specific corridors and counterparty types where conditions are right.

Satoshi wanted to reinvent money. Stablecoins want to finish the job. They will find their place. It will just be narrower than the hype suggests, and more durable than the sceptics admit.

Cross-border payments is not a single business. It is more like dozens of businesses stitched together, each with its own cost structure, its own regulatory requirements, its own competitive dynamics, and its own margins. A company that is profitable on USD-GBP can be losing money on USD-NGN, while charging the same fee for both. Understanding why this happens means going deeper into the economics of a corridor.

What most people believe

The common mental model looks at corridors as interchangeable pipes. Money goes in one end; money comes out of the other. The fees, costs, and margins are roughly the same. So, if you support a corridor, you are in that business. If you don’t, it is a coverage gap. As simple as that.

But this is wrong. Two corridors that look identical on the website (same fee, same promised time, same interface) can have completely different cost structures, operational complexity, and margins. Understanding economics at the corridor level is a prerequisite to fully understanding the company’s business.

The six variables that determine corridor economics

Variable 1: Banking partners

Every cross-border corridor needs at least one banking partner on the receiving end to complete the payment. In well-served corridors such as USD-GBP or USD-EUR, dozens of banks are there to process payments. They compete for volume, which keeps costs low and service levels high.

In a corridor like USD-BDT (US to Bangladesh) or USD-ETB (US to Ethiopia), there may be two or three banks willing to handle the payments. Sometimes even one. When a single banking partner controls your access to a corridor, they are in a dominant position and set the terms. This leads to higher prices, lower service levels, and greater concentration of operational risk. If that partner has an outage, or decides to exit the business, or raises their fees, you have no fallback.

Variable 2: FX liquidity

Different currency pairs trade at different levels of market depth. For example, USD-EUR and USD-GBP are among the most liquid currency pairs in the world. The spread between buy and sell is tight, the cost of conversion is low, and the rates remain competitive throughout the day.

On the other hand, USD-NGN (Nigerian Naira) or USD-BDT (Bangladeshi Taka) are thin markets with fewer counterparties. The spread is wider, the cost of conversion is higher, and the rate can move significantly between the customer initiating the payment and the FX being actually executed. This creates both cost and risk.

In thin FX markets, the payments company often cannot source FX as per the rate quoted to the customer. The margin on this transaction gets compressed by market movement. Companies manage this either by building a buffer with a higher spread or by hedging the currency rate. Both of these mitigants add to the cost of operating in this corridor.

Subscribe now

Variable 3: Regulatory complexity

Every receiving country has its own regulatory requirements. In the UK and Europe, the framework is well-defined and consistently applied. That makes it simple to navigate it with a standard process.

In countries such as India, Brazil or Nigeria, the regulatory landscape is more complex. India has specific rules on the purpose of inbound remittances that vary by amount and recipient type. Brazil’s central bank has specific reporting requirements on cross-border transactions. Nigeria’s FX regulations change frequently, with sudden restrictions on how the Naira can be received or sent.

Each layer of regulatory complexity adds to the operational cost. It requires local compliance expertise and sometimes even a local entity. Such corridors can require months and months of regulatory work before a single payment is processed.

Variable 4: Last-mile infrastructure

Getting money into a country is the first part of the problem. The second part is getting it to the recipient.

In the UK or EU, the last mile is a bank transfer to the recipient’s account. The infrastructure is standardised, fast, and reliable. In emerging markets, the last mile is the hardest part of the entire chain. In Kenya, the most efficient last mile is M-Pesa, not bank transfer. In the Philippines, many recipients want to collect cash through agent networks. In India, the UPI rail has transformed last-mile delivery, but not all payment companies are integrated with it.

Each last-mile method has its own integration requirements, fee structure, and operating processes. Bank deposits, mobile money, and cash pickup mean working with different partners, different settlement cycles, and different operational risks. Companies cannot just add these options to their offering.

The last mile is very critical for the customer experience. If a payment reaches the country in 2 hours but takes 24 hours to reach the recipient’s mobile wallet, it is not a fast payment. The slowest link in the chain determines the speed of the payment. And that link is almost always the last mile.

Variable 5: Competitive landscape

Some corridors are very competitive with multiple providers. For example, USD-INR has a large number of banks and fintechs operating. Competition reduces margins because customers can easily compare and switch. To maintain margins, the payments company needs either a cost advantage or a distribution advantage.

Other corridors have fewer competitors. A company that builds reliable infrastructure for an underserved corridor (say, GBP to GHS or EUR to UGX) can operate with healthier margins because customers have fewer alternatives. But the trade-off is dealing with lower volume, higher operational complexity, and significant investment relative to the revenue generated.

Companies can choose to compete in high-volume, low-margin corridors with well-established infrastructure, or build in higher-margin, low-volume corridors where barriers to entry are high. Companies that do both often end up with margins from niche corridors subsidising losses in competitive ones, which is not a sustainable structure. This is something they need to watch out for.

Variable 6: Volume in the corridor

Corridor economics have a high leverage on costs. Most of the costs (banking partner fees, compliance infrastructure, technology integration, and pre-funding setup) are fixed. As the volume increases, the cost per transaction drops significantly.

A corridor doing 100 transactions per month might cost $10 per transaction to operate. The same corridor, doing 10,000 transactions per month, might cost $2. This creates a chicken-and-egg problem for new corridors: you need volume to make the economics work, but you need competitive pricing to attract volume.

Most payments companies solve this by subsidising new corridors from the margin of established ones. This is the rational choice in the short term. But if the new corridor never reaches the required volume, the subsidy becomes permanent. This leads to payments companies supporting 150 countries, but only 30 of those have profitable unit economics.

Subscribe now

Corridor P&Ls

Each corridor has an implicit (and sometimes explicit) P&L. The revenue side is the total take from that corridor (fees plus FX markup plus float revenue). The cost side includes banking partner fees, FX sourcing costs, allocated compliance overhead, last-mile delivery costs, and the cost of capital for pre-funded accounts.

The margins across corridors can vary widely, even within a company’s portfolio. For a company with flat pricing, a well-established corridor like USD-GBP might have a 60-70% gross margin, while a new or complex corridor like USD-ETB might have zero or even negative margins.

Most payments companies do not measure corridor-level P&Ls. They report blended margins for the corridor portfolio and attribute any changes to “competitive pressure” or “market dynamics.” A good operating practice for payments companies would be to treat each corridor as a separate business having its own margin target, cost-reduction roadmap, and decision criteria for whether to continue operating or exit.

What this means

Corridor economics is the foundation on which all major decisions at a cross-border payments company sit. Pricing, GTM, retention, compliance, and settlement speed: all of these are corridor-specific, not company-wide.

This has three consequences.

First, corridor coverage is an incomplete metric for measuring business effectiveness. The complete picture is: in how many of those countries do you operate profitably, with reliable infrastructure, competitive pricing, and consistent settlement times? Investors and operators who evaluate payments companies on corridor count are measuring the wrong thing.

Second, having corridor depth will beat just having corridor breadth. A company that operates 30 corridors with direct banking relationships, pre-funded accounts, and strong last-mile infrastructure in each will outperform a company that operates 80 corridors through aggregators and correspondent chains.

Third, if you are running a payments company and you do not have corridor-level P&Ls, you do not know which parts of your business are making money and which parts are being subsidised. You are flying blind, relying on blended averages, and averages tend to lie. Companies scale much more profitably when they measure profitability by corridor rather than at the portfolio level.

Every cross-border payments company is a portfolio of corridor businesses. You can manage them as individual businesses, or you can manage them as a blended average. The former approach helps with profitable scaling, while the latter hides the problem until it is too big to fix.

The reason for this misconception is that the biggest fintech success stories of the last decade all had exceptional UX - Wise, Revolut, Nubank, and Chime. It is natural to look at them and conclude that their UX was the competitive advantage. But this confuses correlation with causation. These companies won because of their structural choices and implementation (Wise’s pre-funded local accounts, Revolut’s multi-currency architecture) that happened to come with great UX.

Yes, the real moats in payments are how the underlying infrastructure is built.

Licensing. A payments company with Money Transmitter Licences in 48 US states, FCA authorisation in the UK, and MAS registration in Singapore has spent years of effort and millions of dollars building a regulatory footprint that cannot be replicated quickly. A new entrant cannot ship this in a sprint.

Banking relationships. Cross-border payments require reliable correspondent banking partners in every corridor. These relationships take months to establish and years to build trust, and are governed by varying risk appetites for every corridor. A network across dozens of corridors is an advantage that is not easy to duplicate.

Subscribe now

Pre-funded local accounts. To offer fast settlement, a payments company needs to hold funds in local currency in the destination country. The company with pre-funded accounts in 30 countries settles in hours. The company, without them, relies on correspondent chains that take days to settle.

Integration depth. Once a business customer has integrated through API and built workflows around your data formats, switching is a major operational effort. The deeper the payments company is embedded, the higher the switching cost.

This is also why building a payments product on top of someone else’s infrastructure is not a viable long-term strategy. Using a BaaS partner’s licence, an aggregator’s banking relationships, and a third party’s settlement network helps with a fast launch. But every structural moat belongs to someone else. The fintech controls the UX and the customer relationship, but nothing underneath. The BaaS partner controls the licence, the aggregator controls the banking relationships, and the settlement provider controls the speed.

This means two things. First, the cost structure is permanently higher because every layer includes the partner’s margin. Second, the dependency on these partners is existential for the fintech. If the BaaS provider loses its licence, the aggregator renegotiates the terms, or the settlement partner deprioritises a corridor, the fintech’s business is severely impeded.

The borrowed model works as a quick launch strategy. But every serious fintech has to build its own licensing, its own banking relationships, and its own settlement infrastructure. The sooner this process starts, the better. The longer it is deferred, the more dependent the fintech becomes on the partners.

A beautiful UX on borrowed infrastructure is like building your ambitious storefront on rented land. The landlord can raise the rent, sell the building, or shut it down. And you will have no say in the matter.

But the playbooks for these three approaches are not interchangeable. Each one requires a different team, a different cost structure, a different product architecture, and has a different timeline to profitability. Picking the wrong playbook, or trying to run all three at once, is one of the most common and most expensive mistakes a payments company can make.

What most companies believe

Conventional wisdom tells us that GTM in payments is a sales problem. So you build the product, hire a sales team, and start calling prospects. If the product is good, customers will come. If growth is slow, hire more salespeople. If a particular channel works, double down on it.

This way of thinking treats distribution as something built separately and then attached to the product. This is wrong. In payments, the distribution model shapes the product itself. A product built for marketplace partnerships looks fundamentally different from one built for enterprise sales, and neither of them looks like a self-serve product. The GTM decision is a product decision, and making it late or making it wrong wastes years of effort.

Playbook 1: Platform and marketplace partnerships

This is the model that Payoneer has pioneered. Instead of acquiring individual business customers one by one, the payments company partners with a platform that already has thousands of them. Freelance marketplaces, e-commerce marketplaces, and gig economy platforms. The platform integrates the payments company into its payout flow, and the payments company acquires the platform’s customers at scale.

The economics look very attractive. Customer acquisition cost is close to zero for any incremental customer because the platform delivers them in bulk. A single partnership with a major marketplace can bring 50,000 customers overnight. The marketplace does the distribution for the payments company.

But five things make this playbook harder than it appears-

First, the sales cycle is long, and the conversion rate is low. Signing a platform partnership in payments is a lengthy cycle involving legal review, compliance alignment, API integration, commercial negotiations over revenue share, and often a pilot period. A partnership that starts as a conversation at a conference in January might go live in December.

Second, the customer you acquire through a platform is their customer, not yours. They use you because the marketplace told them to. Their loyalty is to the platform, not to you. If the platform switches providers, the customers leave with it. This means the retention forces described in Memo #2 (integration depth, settlement dependency) operate at the platform level, and not the individual customer level. Losing one marketplace relationship can mean losing thousands of customers at one go.

Third, the margins are thinner by design. The platform expects a revenue share or a preferential rate in exchange for delivering its volume. In a well-negotiated platform deal, the payments company might retain 60-80% of the gross revenue per transaction. In a poorly negotiated one, it could be only 20-30%. This means high volume, but low margin per transaction. This playbook only works at scale. At anything less, it can be painful.

Fourth, you have no control over your customer economics. If your unit economics on a particular corridor or customer segment are not working, you cannot fix it. You cannot decide not to onboard some customers. You cannot reprice the customer directly. You cannot change the onboarding flow to push them toward higher-margin products. Everything flows through the platform, and the platform’s priorities are not the same as yours.

Fifth, the marketplace has an obvious incentive to work with multiple payment providers. Working with a single payout partner makes the marketplace operationally fragile. If that partner’s service is disrupted due to a compliance issue, a banking partner outage, or any other reason, the platform’s payouts stop. So the platform wants to sign a second and a third provider. Now you are competing on price and reliability with other providers in the same account. The relationship that was like a partnership when you signed the deal starts to resemble a competition. And you are at a disadvantage because the marketplace has optionality, while you have a dependency.

The companies that succeed with this playbook share a common trait: they build the product around the platform’s needs, not the end customer’s. The onboarding flow, compliance documentation, payout options, and reporting dashboard - all of it is designed to reduce friction for the platform, because the platform is the real buyer. The end customer’s experience matters only insofar as it reduces the number of support tickets that flow back to the platform.

Subscribe now

Playbook 2: Direct self-serve

This is what Wise has mastered. In this playbook, the customer discovers the product (through word of mouth, search, or advertising), signs up, completes onboarding, and sends a payment. The entire journey is zero-touch self-serve.

The economics of this playbook are different. Customer acquisition cost is higher per customer because each one is acquired individually through marketing spend. But the margin per customer is also higher because there is no marketplace taking a revenue share. And the customer belongs to the payments company, not to a marketplace.

The product requirements for this playbook are quite high. The onboarding has to be fast and frictionless, which means the compliance process needs to be largely automated. Pricing has to be transparent because a self-serve customer has no one to explain the fee structure. The product has to be intuitive enough that a first-time user can complete a transaction without help, which means the complexity of cross-border payments (multi-currency, compliance documentation, variable delivery times) has to be absorbed entirely by the interface.

(This is one reason Wise made the FX markup visible. Transparency is a product requirement of the self-serve model. A customer who has no one to call needs to trust the price on the screen.)

This makes the self-serve playbook expensive to build. Building a product that requires zero touch means investing more in design, automation, and compliance technology upfront. The payoff is that at scale you have a much lower marginal cost per customer. But the upfront investment is significant, and the time to reach scale can be long.

The self-serve playbook also has a natural ceiling in B2B payments. While a freelancer sending $500 can self-serve easily, a mid-market company with $250,000 across four corridors typically cannot. The customer’s internal processes (procurement, legal, treasury) need a human conversation to navigate. This is where many self-serve payments companies hit a wall - they saturate the small-business segment and struggle to move upmarket without adding a sales layer.

Playbook 3: Enterprise sales-led

This is the Adyen model, and also increasingly the Stripe model. Here, a sales team identifies target accounts, runs a consultative sales process with custom pricing, manages the technical integration, and maintains the relationship through dedicated account management.

Here, the economics are inverted compared to the self-serve model. Customer acquisition cost per account is very high. A single enterprise deal can take 6 to 12 months to close and involve solutions engineers, legal teams, and multiple stakeholders. But the revenue per account is also very high. One enterprise client doing $50 million in annual cross-border volume generates more revenue than 10,000 self-serve customers sending $500 each.

This playbook requires a fundamentally different product. Enterprise customers need custom pricing (as covered in Memo #1, the gap between the stated fee and the negotiated rate is widest in enterprise). They need bespoke integration with their existing ERP and treasury systems. They need SLAs on settlement timing, uptime, and support response. They need dedicated compliance handling for their specific industries. All this means the product cannot be standardised the way a self-serve product can be. It has to be configurable, and configuration requires human touch.

The risk of this playbook is the pricing death spiral described in Memo #3. Enterprise customers are acquired at custom rates that assume volume growth. When the growth does not materialise, the rate becomes the permanent rate. When the company tries to adjust, the customer threatens to leave. The discounting cycle begins. This spiral is common in enterprise sales because each deal is a unique negotiation, but ends up setting a precedent.

The other risk is concentration. A payments company with 30% of revenue coming from three enterprise clients is sitting on huge concentration risk. If any one of these clients renegotiates its price or churns, the revenue impact is immediate and significant.

The companies that succeed with this playbook share three characteristics -

First, they treat pricing as a system, and not a series of negotiations. They build structured pricing tiers with clear rules for custom rates, and they have the discipline to walk away from deals where the margin does not work. As covered in Memo #3, this discipline is hard to implement. Adyen is known for not chasing price-sensitive merchants. They are rarely the cheapest option in a competitive process, and they do not discount aggressively to win. This means they lose some deals. But the deals they win are at healthy margins with customers who value reliability over cost.

Second, they go deep in specific verticals rather than selling horizontally to anyone who takes their meeting. Adyen built vertical expertise in travel, retail, marketplace platforms, and digital goods. Their sales team understands chargeback patterns in travel, the omnichannel requirements of retail, and marketplace payout complexity. This leads to a sales conversation that a generalist competitor cannot have, and it builds a product that gets better for each vertical over time.

Third, they focus their account management teams on technical after-sales rather than on relationships. They anchor the conversation on optimising authorisation rates, reducing declines, and improving checkout conversion. These are product conversations, not pricing conversations. When the value delivered is tangible, the cost of switching becomes high.

Subscribe now

Most companies try three and fail at two

Most payments companies, especially ones between Series A and Series C, try to run two or three of these playbooks simultaneously. They have a partnerships team chasing platform deals, a marketing team running self-serve acquisition, and a sales team closing enterprise accounts.

This rarely works. Each playbook requires a product built around it. The self-serve product needs zero-touch onboarding and transparent pricing. The enterprise product needs configurability and custom pricing. The platform product needs to be built around the platform’s API and operational requirements. Building a single product that serves all three is technically possible, but extremely difficult. It almost always results in a product that serves none of them well.

But the constraint is beyond just product. Operations, compliance and licensing - all have to be designed around the chosen playbook. A self-serve model requires automated KYC and compliance flows that can support thousands of applications without human review. An enterprise model requires customised compliance handling, with dedicated teams that can navigate a client’s specific regulatory requirements across jurisdictions. A platform model requires compliance frameworks that work at the platform level, not the individual customer level, which is a fundamentally different architecture.

The same applies to the entire operations - settlement cycles, customer support, escalation workflows, and treasury management. Each looks different depending on whether you are serving 100,000 self-serve customers, 500 enterprise accounts, or 5 platform partnerships. A support team built for high-volume, low-touch self-serve queries cannot handle the complex, multi-stakeholder escalations that enterprise clients expect. A treasury team optimised for pre-funding a handful of corridors for platform payouts cannot manage the corridor diversity that a self-serve model with global customers demands.

When a company tries to run all three playbooks, it is not just the product that gets stretched. It stretches the entire operating model. It could end up with compliance processes that are too manual for self-serve and too generic for enterprise, or operations SLAs that are too slow for platforms and too expensive for small customers, or licensing decisions that are made to support one playbook, but which create constraints for another.

This also creates an organisational problem. The partnerships team, the sales team, and the marketing team each have different incentive structures and different definitions of success. Their incentives and definitions of success can, and invariably will, conflict. A self-serve customer acquired through marketing at a $50 CAC with transparent pricing does not want to learn that an enterprise customer with the same volume pays half the rate because they negotiated.

What this means

The GTM decision in payments is not a simple go-to-market decision. It is a strategic decision that has bearing on the product, the org structure, the pricing model, and the unit economics.

This means three things for payments companies -

First, the playbook you choose constrains your entire organisation, not just your product. A self-serve product cannot be retrofitted for enterprise without significant investment in engineering, compliance, operations, and customer support. An enterprise operation cannot be simplified into a self-serve experience by removing features and people. These are not just adjustments - they need rebuilding across every function.

Second, the unit economics of each playbook are fundamentally different. Comparing payments companies across playbooks using the same metrics is misleading. A platform-partnership company with $10 billion in volume and thin margins is not comparable to a self-serve company with $1 billion in volume and high margins. They are both playing different games with different scorecards.

Third, “we’ll do all three” is not a strategy. It is an indirect admission that the company has not yet made the GTM decision. And the longer this decision is deferred, the more expensive it becomes. The product, compliance, operations, organisation structure, and pricing structure continue to accumulate debt as they serve multiple models simultaneously.

Every payments company has a GTM strategy. Yet, most of them end up running three. And that is the problem.

Square did it. Stripe did it. PayPal did it. Razorpay did it. Adyen did it. And the growth rates for their capital product and the repeat borrowers for the capital product are clear signs that these are not experiments but full-fledged product lines.

Why is this happening?

A mature payments company has the customer fully onboarded, can see their cash flow data, controls the settlement pipe (which can also act as a collection pipe), and has the customer’s attention. A bank evaluating the same customer relies on months-old financial statements and a credit score that is far removed from the recent business health. So, the payments company can observe revenue in real time, verify it against actual transactions, and deduct repayment from the next payment.

Every major cost line in traditional lending is cheaper for a payments company lending to its own customers.

There is also a margin argument. As covered in Memo #3, payments margins face downward pressure. A company doing $10 billion in annual volume at 20 basis points makes $20 million. The same company lends only $500 million at a 4% net interest margin to make the same $20 million.

For any payments company watching its spreads narrow, this maths is difficult to ignore. This is a logical product to improve the margin. And there is a pressing customer demand for capital as well.

So the economics work, there is a real data advantage, and the customers need the product. It’s a great fit and easy to build.

This complacency leads most payments companies to underinvest. They assume that because they can underwrite better and distribute cheaper, the hard part is done. But lending requires credit risk management through the life of the loan, not just at origination. It requires sourcing funds beyond what’s on the balance sheet, whether through banking partners, credit facilities, or securitisation. It requires a real collections function for when the borrower’s business deteriorates. And it requires continuous investment in improving the underwriting model, because the first-generation model built on transaction data will not hold up through a credit cycle without iteration.

Subscribe now

Most payments companies don’t treat these as core capabilities that require investment. They understaff the risk function, defer the fund sourcing to their payments treasury, don’t build the collections infrastructure, and move the engineering team back to payments features right after the lending product launch. The result: strong numbers in year one, rising delinquencies in year two, and a difficult conversation with the board in year three.

Square (now Block) is the best example of a company that succeeds by investing in these capabilities. It has underwritten more than $22 billion in loans, with aggregate loss rates below 3% - without conducting any traditional credit checks or asking for tax returns. To put this in context, the Equifax Small Business Default Index for traditional bank lending, in which borrowers undergo full credit checks, is around 3.2%. Square is matching the benchmark without any of these traditional checks, relying almost entirely on transaction data.

This loss rate is not a fluke. Square built repayment capabilities by wrapping it into its settlement flow, maintained origination discipline to onboard high-quality customers, and invested in the lending operation as a separate capability. The result: sellers who take a loan use an average of 3.7 Square products, versus 1.5 for those who do not, and retention improved by 15% for these sellers. The lending business, besides generating revenue, also deepened the payments relationship.

Every payments company faces this decision when they reach sufficient scale. The economics are too attractive, and the customer needs are too obvious to ignore. The real question is whether they are willing to invest in risk, collections, fund sourcing, and underwriting with the same seriousness as they do in payments.

The payments infrastructure gives you the distribution. It does not make you disciplined. And in lending, discipline decides whether you make revenue or incur losses.

This story is so accurate that it’s boring. So let’s talk about what the press release does not say.

Four things the press release does not say

One. The premium depends on whatever date you pick.

Calcalist quoted a premium of 21% to Payoneer’s last close before the deal was confirmed. The Globe and Mail quoted 10% for close on that same Friday - a different number for an identical reference point, most likely because they used different assumptions about Payoneer’s share count.

The Globe and Mail and Yahoo Finance Canada separately put the premium at 40%, using the price from the week before Reuters first reported the talks, reflecting a completely different anchor.

One more data point: the Global X FinTech ETF, a broad fintech benchmark, is down roughly 18% so far in 2026.

There are divergent data points. Determining the premium paid in this deal is genuinely hard; no single number gives an objective view of what it actually is.

Two. The float is as important as the business.

Payoneer’s adjusted EBITDA for 2025 was roughly $270 million. Of this, more than $200 million came from interest income on customer balances of nearly $8 billion. The ex-interest EBITDA was closer to $40 million. So, Nuvei could be buying a payments company with a healthy balance sheet, or it could be buying a healthy balance sheet with a payments company attached.

Of course, Payoneer’s core business is improving fast. The 2026 guidance for Payoneer filed with the SEC shows ex-interest EBITDA more than doubling to $85-95 million. But remember, 34% of 2025 revenue came from Greater China, according to Payoneer’s own filings. This kind of concentration is riskier than usual today, given the macroeconomic environment and the tariff uncertainty. The risk of falling interest rates can be modelled and priced into a deal, whereas the risk of tariff changes affecting Greater China business cannot.

Three. The two go-to-market engines barely overlap. But the two headcounts do.

Nuvei sells acceptance to enterprise merchants with a long, technical, sales-led process. Payoneer acquires customers through marketplace partnerships, self-serve, and direct sales. The two GTM engines are designed for different products and buyers, and both companies have signalled they intend to run them separately rather than force one onto the other.

However, the part nobody is saying out loud is about the overlap in headcount. Payoneer employs roughly 2,540 people, and about 51% of them are based in Israel, including most of its product and engineering teams. Nuvei, too, has a deep Israeli footprint through its acquisitions of SafeCharge and Simplex. This overlap raises not just a cost-optimisation question but also an integration question. The people doing redundant-looking jobs in Tel Aviv may be the same people whose expertise keeps Payoneer’s licences in China and India running. Cutting the wrong roles during integration could cost Nuvei the very licences it has paid for.

Add to this a PE owner accustomed to cutting costs, integrating a public company with a more relaxed, employee-positive culture, and a deal close date more than a year away. That is a long runway for exactly the people Nuvei cannot afford to lose to leave first.

Subscribe now

Four. The real prize might not be in the press release.

In February, Payoneer applied to the US OCC for a PAYO Digital Bank charter. This national trust bank would allow it to issue its own dollar stablecoin and hold the reserves backing it. Payoneer’s own executives have admitted publicly that the company’s gap is the last mile, i.e. turning a stablecoin balance into usable local currency across the markets where its customers operate. Nuvei’s 2021 acquisition of Simplex already gave it exactly that: fiat-to-crypto on- and off-ramp infrastructure, plus an e-money licence in Europe. Put the two together, and the combined entity gets a complete loop for stablecoins: issue, hold, convert, spend.

Stablecoins are still small, roughly 1% of global FX flows. But according to Standard Chartered and Zodia Markets, there is a path to grow to 10% as regulation matures. Stripe also made a comparable bet on optionality in 2024, paying $1.1 billion for Bridge despite its small revenue base at the time. So, if this bet plays out for Nuvei in any reasonable timeline, it will add more to this deal’s long-term value than anything in the cost optimisation plan.

What it means

When we add these up, this deal is not really about acceptance meeting payouts. Nuvei is making three separate bets: (1) that a balance sheet heavy with interest income can convert into real payments margin before falling rates remove the cushion, (2) that two different go-to-market engines and two overlapping workforces can be merged without losing the people critical to operating in the licensed geographies, and (3) that stablecoin settlement grows fast enough to matter.

All three are big bets for the future. How these play out will determine whether the cheque paid for the deal was worth it.

Some companies have found structurally cheaper ways to move money. Better banking integrations, more efficient treasury models, lower-cost infrastructure. Their prices are lower because their costs are lower. This leads to a permanent market shift, and it is healthy. This memo doesn’t cover this scenario.

The second is when a payments company prices aggressively to win a customer. The strategy works, and the customer comes on board. As the relationship progresses, the margin on the customer comes under strain. The company tries to widen the FX markup or increase the fees to maintain its margin. The customer notices, shops around, and threatens to leave. The company offers a discount to keep them. The margin gets worse. The customer has learned that threatening to leave gets them a better deal.

This is the pricing death spiral. When a company is in this cycle, every step feels rational when taken individually. But collectively, it degrades margins, trains customers that behaving adversarially is better for them, and leaks competitive intelligence. Most payments companies inevitably get caught in this cycle now and do not recognise it as a pattern.

How the spiral starts

The spiral always begins with a rational decision: win the customer.

Suppose the payments company is competing for a mid-market customer doing $2 million per month in cross-border volume across three corridors. The customer is evaluating between two providers. The sales team knows the standard rate card will not win the deal, because the competitor is quoting lower. So the company offers a custom rate: a reduced stated fee, a tighter FX spread, or both. They win the deal.

This is a commercial win. But two things that will matter later have happened in parallel.

First, the customer was acquired at a margin that assumes their volume will grow. The pricing was justified internally with a note that says something like, “at $3 million monthly volume, this customer becomes profitable at these rates.” So, the customer’s current volume does not support the margin. And the future volume is a projection, not a commitment.

And projections, more often than not, do not materialise. Six months later, the customer is still at $2 million. The company now has a choice: either enforce the pricing terms contingent on the volume commitment or accept the current rate and absorb the margin gap. In theory, the company should enforce. In practice, no company actually does. Losing $2 million in monthly volume to enforce a pricing condition is a decision no commercial team wants to make. So the unsustainable rate becomes the permanent rate.

Second, the customer now has an anchor price. They believe this is what the service costs. Any future increase will be measured against this number, regardless of the growth assumption it baked in or whether it was sustainable for the company.

How it escalates

Six to twelve months pass. Now the customer’s volume has grown, but not to the projected level. The margin on the account is thin or negative. The company’s finance team flags it. The pricing committee reviews the account and decides to adjust the rate.

They make a small adjustment – maybe a modest increase in fees, or a few basis-point increase in the FX spread. They expect the customer will absorb this increase without noticing or complaining, because the new rate is still very competitive.

But the customer does not evaluate this new rate in absolute terms. They evaluate it relative to the anchor. Let’s say they were paying 0.1% fees. Now it is 0.15%. That is a 50% increase in their FX cost, even though the absolute difference is just 5 bps. This increase versus the anchor is what they discuss internally.

The customer starts looking for other options. They are now very aware of their pricing, which they weren’t earlier. The price increase, which was meant to fix a margin problem, has created a retention problem.

The discount trap

The customer calls their account manager. They have received a quote from a competitor. The quote may be real or fabricated, but it does not matter. The threat of losing the customer is now on the table.

The account manager escalates to get a pricing exception. The company now faces a choice: hold the new rate and risk losing the customer, or offer a retention discount and keep the volume.

Most payments companies choose the discount. The logic is simple: compare the cost of acquiring a new customer to replace this volume to the cost of the discount. The unit economics, evaluated in isolation, favour retention.

But the discount does five things that the unit economics do not capture.

First, it confirms to the customer that the original price increase was negotiable. Every future increase will now be met with the same behaviour: threaten to leave, wait for the discount. The company has trained the customer to negotiate using threats.

Second, it signals to the customer that the company’s stated pricing is not its real pricing. The customer now knows there is a gap between the rate card and what the company is willing to accept. This is the same transparency problem discussed in Memo #1, but created by the company itself.

Third, it leaks pricing intelligence to competitors. The customer can use the retention discount to further negotiate with a competitor. “This provider is offering 0.1% rate plus a cashback. Can you match this?” The competitor now knows the company’s floor price and can use it in future deals.

Fourth, it creates an internal precedent. The next time any account manager faces a similar situation, they will also expect this exception. This makes discounting the default response to any churn risk.

Fifth, the discount does not stay private. This is less understood by most companies, and they treat each pricing deal as a confidential concession to that customer. But in concentrated verticals like e-commerce or SaaS platforms, customers are a small, tight-knit community. They talk to each other, meet at industry events, and are part of the same WhatsApp groups. When one customer gets a discount, others will find out. They then follow the same playbook: threaten to leave, provide the new quote and get a discount. The company has set a new price expectation for the segment.

The structural damage

This spiral, once it catches on, spreads through the portfolio from both ends, i.e., new customer acquisition and existing customer retention.

On the retention side, the pricing floor keeps dropping. Each discount becomes the new reference point for the next negotiation. Customers who were never planning to leave start asking for pricing reviews because they have heard that others received better terms. The account management team spends more and more time on pricing discussions and less time on growth conversations.

On the acquisition side, the sales team is acquiring new customers at increasingly aggressive rates due to the tight competitive environment. The new customers come in at lower margins than the existing ones, which means the portfolio’s blended margin is declining from both directions: existing customers negotiating down and new customers being acquired cheaply.

When the finance team sees this drop in margin, they attribute it to “competitive pressure.” This is only partially correct. The competitive pressure is a cause, but the company’s response to it is accelerating the damage.

Why is the spiral hard to stop

The pricing death spiral is self-reinforcing because each step in the cycle creates the conditions that aid the next. Aggressive pricing to win a deal means thin margins. Thin margins force price adjustments eventually. Price adjustments trigger customer awareness. Awareness leads to shopping. Shopping leads to churn risk. Churn risk leads to discounts. Discounts reduce the margins further.

Breaking out of this cycle is difficult because it requires a company to accept short-term pain for long-term structural health. Specifically, it requires three things that most payments companies resist:

The first is walking away from deals that require unsustainable pricing. If a customer can only be won at a margin that depends on projected volume growth to be sustainable, the company is not pricing the current deal, but a forecast. And as described above, when the forecast does not materialise, the company never enforces the original terms. The temporary rate becomes permanent.

The second is holding the price when a customer threatens to leave. It is hard to maintain discipline here. The immediate loss of volume is visible. The structural benefit of pricing discipline is invisible and long-term. Every payments company knows this, but ends up making frequent exceptions.

The third is separating acquisition pricing models from retention. Most payments companies use the same pricing logic for both. The rate at which a customer was acquired becomes their forever rate. However, building pricing logic with a clear, well-communicated path from introductory to standard rates, with transparent triggers for the customer at each step, removes the fuel for the spiral.

When the market itself shifts

This is where the company’s own decisions drive a spiral. There is a worse version: when the market itself reprices structurally.

This happens when a competitor finds a way to operate at permanently lower costs. Maybe they have built direct integrations with local banks that bypass correspondent banking fees. Maybe they have a treasury model that allows them to pre-fund corridors more cheaply. Maybe they are a new entrant with a different cost structure, funded by venture capital that is willing to subsidise growth, or built on infrastructure that is genuinely cheaper to run.

In these cases, the competitor is not offering a tactical discount to win a deal. They are operating at a price point that the incumbent cannot match without restructuring its own costs. This is a structural repricing of the market, fundamentally changing market dynamics.

When the competitive threat is tactical (a one-off discount to win a customer), the company can hold its price and take a bet that the competitor will not continue to play the discounting game. When the competitive threat is structural (a permanently lower cost base), holding price means losing customers steadily, and matching price means operating on margins the business was not built to support.

When the shift is structural, it becomes an existential problem. The company is now operating in a market with a price point that compresses its margins to unsustainable levels. The playbook that worked for tactical competitive pressure (hold price, add value, deepen integration) does not work after the market has repriced.

At this point, the company needs to either restructure its costs or find a way to justify a premium through product differentiation. Neither of these is quick or easy.

What this means for the company

The pricing death spiral is a structural problem that shows up in pricing decisions. It can be avoided by thinking about it structurally.

The companies that successfully avoid it share three characteristics. First, they build enough product stickiness (through integration depth, settlement dependency, and corridor coverage, as discussed in Memo #2) that makes switching expensive for the customer. When switching is expensive, the customer’s decision is no longer driven by who offers the lowest rate. The company can price based on the value it delivers, rather than reacting to any number a competitor puts on the table.

Second, they treat pricing as a product decision, governed by rules and architecture, rather than a commercial decision, governed by negotiation and discretion. When pricing sits with the sales team, every deal becomes a one-off negotiation. When pricing sits with the product team, it becomes a system with logic that holds across customers.

Third, they have the discipline to let customers leave when the economics do not work. This is the hardest of the three, but it is important for escaping the spiral.

The companies stuck in the spiral share a different set of characteristics. Their sales team has more pricing power than their product team. Their retention strategy almost always is a discount. And their margin compression is explained as ‘market dynamics’.

A payments company that cannot spell out, for any given customer, the margin at acquisition, the margin today, and the trajectory of that margin over the next twelve months does not have a pricing strategy. They are conflating pricing strategy with a series of negotiations. And negotiations, without a structure, tend to spiral.

A business is looking for payment providers. They check the fees listed on the pricing page and compare two or three alternatives. They find the option they find competitive, and sign up. They use the product, and it works well. Things are fine for six months, or maybe a year.

Then something changes. Maybe a new FP&A hire runs a reconciliation. Maybe the Finance Head looks at a quarter’s transactions and compares the FX rates they received with the mid-market rates for those days. They add up the spread and factor in the settlement time. They work out what the company actually paid per transaction, all in.

This number is not what they thought they were paying. The stated fee was 0.5%. The real cost, after including the FX markup and the float, is closer to 1.5%. At a million dollars in monthly volume, that is a difference of $120,000 per year.

This customer is now lost. This does not mean they stop using the product immediately. That may take a month or a quarter, but the customer has been lost. The company will mark it as churn, diagnose it as a retention problem, and respond with the usual playbook: assign an account manager, offer a discount, and schedule a monthly call for review. But none of it will work.

It will not work because the customer is not leaving over price. They are leaving because they feel they were misled. These are two completely different issues, and they require completely different responses.

A price problem is rational. The customer has done the comparison, found someone cheaper, and is making an economic decision. You can fight this with a better offer. You can add value somewhere else. The customer is still at the table, having a commercial discussion with you.

A trust problem is different. The customer believed something to be true, but discovered it was not. They are now re-evaluating everything you have ever told them. The account manager is no longer a partner; they are a representative of the company that misled them. The monthly meeting is no longer a growth or collaborative conversation; it is a confrontation.

The payments industry has a structural reason for this pattern. Payment pricing comprises five components stacked together: the stated fee, the FX markup, the float, the corridor cost, and interchange (as we discussed in Memo #1). The customer sees the first one, while the rest are hidden. The longer a customer pays the hidden price without knowing it, the larger the gap becomes between what they believed and what was true.

Wise has proven that transparency in consumer payments can give you a competitive edge. They made their FX markup visible. They showed customers the mid-market rate and their markup as a separate line item. This single design decision turned an opaque industry into one where consumers could clearly see what they were actually paying.

Once customers experience that level of transparency, they do not go back. The expectation moves in only one direction – towards higher transparency. And the business customers making cross-border payments are the same people who use Wise and Revolut in their personal lives. They already know what transparency looks like. They will eventually demand it from their B2B providers too.

The B2B market has not had its Wise moment. Business customers are still discovering their real costs through reconciliation, not through the product. The company that changes this will build a moat that is very difficult to breach. A customer who already knows exactly what they pay has no gap left to discover. There is no surprise discovery to be made.

If you are running a B2B payments company and your churn analysis says that customers are leaving for better pricing, go back and look at the data more carefully. Ask when the customer started the switching process. Ask what happened in the weeks before. Most likely, you will find a moment of discovery that led someone to ask a pricing question no one had asked before.

The companies still hiding the FX spread will keep losing customers and keep calling it a retention problem. It is the inevitable consequence of a pricing structure that depends on customers not doing the maths. The churn is not a customer success problem; it is a transparency debt. And like all debts, it has to be eventually paid.

The common assumption made is that customers stay because they like the product and leave because they find a cheaper alternative. This logic leads payment companies to respond to customers about to churn with discounts, and to loyal customers with neglect. These responses are wrong because the underlying mental model of why customers stay or leave is flawed.

Payment retention does not work like SaaS retention. In SaaS, the product is the experience. If the experience is good, the customer renews. In payments, the product is invisible when it works. The better the product, the more invisible it is. The customer does not think about the payments provider until an issue arises. This means the forces that hold a customer in place are largely structural, rather than experiential. And the forces that push a customer out are largely triggered by specific events and not by gradual dissatisfaction with the product.

Understanding how the retention forces and churn triggers work is necessary to build a payments company that retains customers by design and does not rely on accidental (and temporary) retention.

Why customers stay

Five forces make a payments customer keep using the product. They are not equally effective or equally easy to build. Most payments companies end up relying too much on the weakest ones, which are also the easiest to build. Listing these below in the order of effectiveness -

Force 1: Integrations

This is the strongest retention force in payments, and it has nothing to do with how good the product is.

When a business integrates a payments provider through an API, it embeds that provider into its core financial infrastructure. The accounting system mappings, reporting dashboards, and reconciliation logic are all built around the data formats of the payments provider. Over time, this becomes too costly to switch, as it would mean rebuilding it all.

The switching cost is not the technical effort of plugging in a new API. That can be done in a matter of days. The switching cost is the operational overhauling needed - remapping accounting codes, rewriting reconciliation scripts, retraining the finance team, and managing the transition period with two parallel systems. This is why mid-market and enterprise payments companies with deep API integrations often have retention rates well above 95%, even when their pricing is not the most competitive.

Force 2: Settlement dependency

Businesses build their cash flow management around their payments provider’s settlement cycle. For example, a company knows that its USD collections settle in T+1 and its INR payouts settle in T+2, so it plans its working capital, supplier payments, and treasury operations around those timelines.

Switching to a new provider with different settlement timelines means adjusting the entire cash flow model. Even when you have a provider who can settle a day faster, you need to adjust all downstream processes affected by the change. And if the new provider settles a day slower, it can throw the cash cycle off and create a big cash crunch.

Force 3: Corridor coverage

Most payments companies do not cover every corridor a business needs. But once a customer has found a provider that meets their needs for a specific combination of corridors, finding a replacement means being sure the new provider can do the same.

This is not a trivial exercise. The more corridors a customer uses, the harder it is to find a single replacement. A logistics company sending payments to drivers in 12 countries needs a provider that supports all 12 destination currencies with reliable last-mile delivery in each. The new provider might cover 10 of the 12 and claim the remaining two are “coming soon.” The customer’s choice is to run two providers in parallel or wait. Most of them choose to wait.

Force 4: Compliance and onboarding

Every payments provider requires KYC and KYB documentation. For a business customer, this means collecting and submitting corporate documents, UBO records, proof of business activity, and sometimes even audited financial statements. This is a strong deterrent, especially for customers in complex industries or high-risk categories.

Once a customer is approved and is transacting, they don’t want to keep doing this verification with new providers. The deterrents are the time this verification process consumes and the risk that the new provider’s compliance team might reject them or impose restrictions that the current provider does not.

Force 5: Relationship and account management

This is the retention force that payments companies invest the most in, and the one that matters the least.

Account managers build relationships with customers, negotiate custom pricing, handle escalations and conduct periodic business reviews. While these may sound like retention activities, if a customer has decided to leave for structural reasons (such as FX markup or settlement delay in a corridor), these activities cannot stop them. A good account manager will delay the churn, but will not be able to prevent it.

Another limitation is that these activities are personal and not structural. When the account manager leaves the company, the retention efficacy leaves with them.

Why customers leave

The five forces above are what keep a customer in place. But they are not permanent. There are specific trigger events that can shake this stability and push these customers from stable to lost. This almost always happens suddenly. A customer who looked stable last quarter is switching providers this quarter, and the payments company never saw it coming.

Trigger 1: Pricing discovery or broken trust

This is the most common trigger for churn in payments. As covered in Memo #1, payment pricing comprises five components stacked together: the stated fee, the FX markup, the float, the corridor cost, and interchange. Most customers only see one of them. When the customer calculates the total cost of their payments, the stated fees plus the FX markup plus the cost of float, they feel misled.

This is why churn conversations that start with “we found a better rate” are misleading. The rate is not the cause of churn. The cause is that they discovered that the rate they were promised was not the true rate. This is a trust problem, and not a price problem.

Trigger 2: The failed transaction

In payments, reliability is a core product feature. When a payment fails, the damage it does is beyond that failed transaction. It can cascade into a supplier not being paid on time, a payroll cycle being delayed, a contract penalty being charged for missed payment, or a business relationship becoming strained.

A one-off failed transaction in a low-stakes context is forgiven. But a failed transaction that causes a real-world consequence for the customer’s business, or multiple failures over time, will trigger the switching process. These failures force customers to evaluate the risk of failure, which they had not factored into their decision on the payments provider.

This trigger is dangerous because it hits suddenly, and by then, the customer has already made up half their mind to leave. They may have even started with another provider before the payments company finds out.

Trigger 3: The compliance friction

Compliance holds are business as usual in payments. But how a company handles them determines whether the customer stays or leaves. A company that communicates proactively, resolves issues quickly, and explains clearly can maintain the relationship and keep the customer. A company that goes silent, sends templatised emails, and takes two weeks to release funds will lose that customer.

The compliance friction itself is not the trigger here. It’s how the company handles the compliance friction. Customers accept that compliance exists. But they do not want to be treated as criminals or suspects by their own payments provider.

Trigger 4: The coverage gap

If a customer’s business grows into a new corridor that the provider does not support, they need a second provider. Once they have a second provider, it’s natural that they will compare and evaluate both providers. Now, the retention forces at play, which were working in favour of the first provider, start to weaken.

The first provider also doesn’t realise this is happening until the volume starts shifting. The customer did not leave because they were unhappy with something. They left when their business outgrew their provider’s coverage, and the second provider turned out to be better than the first.

What this means

The retention forces and the churn triggers are fundamentally different. The retention forces are structural and slow. The churn triggers are events and operate fast. This is why payments companies are consistently surprised by churn. A customer appears stable for years because retention forces hold them, only for a single trigger event to overwhelm them. The customer is then gone in weeks.

This has three consequences for how payments companies should think about retention.

First, the strongest retention is built into the product, not around it. Integration depth, settlement dependency, and corridor coverage are product decisions. They are not hooks that the Customer Success team can create. The companies with the highest retention are the ones that have designed stickiness into their product.

Second, monitoring the trigger events matters more than measuring NPS or CSAT. A customer who reports high satisfaction can still leave next month if they discover their true cost or suffer a failed payment at the wrong time. The payments companies that successfully reduce churn are the ones that carefully monitor the triggers: compliance hold frequency, unusual reconciliation activity, and corridor coverage gaps relative to the customer’s growth.

Third, the weakest link in most retention strategies is the overreliance on account management to do what the product should be doing. If a customer needs an account manager to get a rate adjustment, to understand their fee, or to resolve a compliance review, the product has failed to absorb the complexity. The account manager is at best a temporary patch, and patches don’t work forever.

Every payments company has customers who look stable right now. But they are one event away from churn. The companies measuring the wrong things, like NPS and CSAT instead of trigger events, or investing in account managers instead of product stickiness, will continue to be surprised by the customers who were stable last quarter, and ‘suddenly’ left.

The Consumer Payments world figured this out a decade ago. Wise does not demo how to make a transfer. Revolut does not walk you through a screen share to open an account. Pix does not require a sales call. You open the app, you do your thing, and you are done. The standard for consumer cross-border payments is zero-touch onboarding and self-serve everything.

It seems that the B2B payment players have missed this trick. They have normalised a workflow in which a business customer fills out a form, waits for a call, sits through a product walkthrough, asks obvious questions that the interface itself should have answered, and then, maybe, starts using the product. Entire sales organisations have been built to compensate for products that cannot explain themselves.

We have convinced ourselves that this is the way to do things. Why? Because “B2B is complex,” and “our customers need hand-holding,” and “the compliance requirements make self-serve impossible.” These are less explanations and more excuses dressed up as industry wisdom.

I do appreciate that B2B payments involve more complexity than consumer payments. Multi-currency accounts, batch payments, approval workflows, and compliance documentation – all are complex. But complexity should not be a justification for bad design. It should be the reason to invest more in good design. The harder the underlying process, the more the interface needs to absorb that complexity. That is what product design is for.

When a payments company requires a demo, what they are really saying is that they have built based on what they understand, and not based on what the customer needs. They have designed their flows based on their internal architecture rather than the user’s behaviour. And now they have hired people to translate it for customers. The demo is this translation.

The cost of this is much more than bad UX. Every demo delays the sales cycle from minutes to days. Every sales call adds to the CAC. Down the line, these customers also need more support tickets, account management, and handholding throughout their lifetime, because the product never taught them to be independent. The dependency, which starts with the demo, stays long enough to impact the LTV.

The companies that will win in B2B payments are the ones building products with zero-touch; that a finance executive can sign up for, configure, and send a first payment through - without needing to speak to a human.

Needing human support to complete a basic workflow means the product has failed at its primary job. The need for demos is not a ‘feature’ or ‘USP,’ but rather an outstanding product and design debt.

This is the paradox of pricing payments. These numbers are all different. Understanding the gap between them is the most important thing to learn about how the payments industry works.

What People Believe

The common mental model for pricing tells us that the payment company charges a fee to send money, and that fee covers their costs plus a margin. If companies offer low prices, they must be more efficient. The expensive companies must be taking advantage of you. Simple.

This is wrong in almost every respect. Looking at payment pricing as a fee is incorrect. It is a combination of interlocking components, each controlled by a different actor, each with its own logic, and most of them invisible to the customer.

The five components of Payment Pricing

Component 1: The Stated Fee

This is the number mentioned on the website. “Send money for $15,” or “1% flat fee.” This is the price the customer ‘thinks’ they are paying.

The stated fee exists for one reason: to give the customer something to compare. When a customer compares three providers, they compare the stated fees. It’s easy to compare 1%, 1.5%, and 3%. While customers think they are rationally comparing costs, they are comparing the wrong numbers.

The stated fee is at best a marketing instrument, with very little to do with the payment company’s revenue model.

Component 2: The FX markup

Let’s say you send $1,000 from the US to India. This payment has to be converted from USD to INR. There is a ‘mid-market rate’- the rate at which currencies trade on wholesale markets. And then there is the rate the payment company gives you.

The gap between these two rates is the FX markup. This markup is never disclosed as a separate line item. This is where the payment company makes real money.

In a real-world example, on a given day, the mid-market rate for USD-INR might be 93. A bank might offer you 91.15, which is a 2% markup. On a $1,000 transfer, you lose $20 - without charging any fee. A company like Wise might offer you 92.75, or a 0.25% markup. That is $2.50.

The difference between these two is $17.5, while the stated fee difference may be as little as $3 on this transaction.

This is why comparing the stated fees is wrong. The FX markup can be 2 to 10 times bigger than the stated fee, depending on the provider and the corridor. For a payments company, FX markup is a primary source of revenue. And yet, this is the component that most customers never check.

Component 3: The float

When you initiate a cross-border payment, the funds are debited from your account almost immediately. But they do not arrive in the recipient’s account for 1 to 3 days, and sometimes even longer. During this time, the payments company (or one of the intermediaries in the chain) holds your money.

The sitting money earns interest. And across a portfolio of millions of transactions, the interest is significant.

This is the float. It is not disclosed to customers. But it contributes meaningfully to the company’s revenue, especially in a high-interest-rate environment.

The float also creates a perverse incentive for payment companies. Faster payments mean less float revenue. When a company says it is investing in speed, it also means it will cut a line item from its revenue to improve the customer experience. While some do this, many don’t, succumbing to revenue pressure.

Component 4: The corridor cost

Money movement across different corridors costs the payment company differently. $1,000 from the US to the UK costs far less than sending $1,000 from the US to Vietnam. Costs depend on infrastructure, banking partner fees, compliance requirements, and FX liquidity, and these differ dramatically by corridor.

In a high-volume and well-regulated corridor like USD-GBP, the cost to the payments company could range from $2 to $4 per transaction. Pre-funded local accounts in the UK, deep GBP liquidity, standardised compliance, multiple banking partners competing for volume. This is a cheap corridor to operate.

In a low-volume, complex corridor like USD-VND, the cost might be $8 to $12. Limited banking partners to handle Vietnam-bound flows, limited FX liquidity that widens the spread and complex compliance requirements for the receiving country. This is an expensive corridor.

And yet, customers in both corridors might see a similar stated fee. The difference between the real cost to the payment company and the stated fees gets absorbed into the FX markup. Simply said, the corridor with the higher transfer cost gets a wider spread. Thus, the customers sending money to Vietnam are bearing a margin they cannot even see.

Component 5: The interchange and network fees (for card payments)

If the payment is made with a credit or debit card, there are additional interchange and network fees.

Every card transaction involves a fee paid by the merchant’s bank (the acquirer) to the customer’s bank (the issuer). This is an interchange set by the card networks. It is not negotiable at the individual transaction level. It ranges from 0.5% to 3.5%, depending on factors such as card type, merchant category, geography, and whether the card is present or not.

In addition to interchange, the card network charges a fee for using its rails. And then the acquirer adds its own margin on top.

So when a merchant sees “2.9% + $0.30” from their payment processor, the breakdown might be 1.8% interchange to the cardholder’s bank, 0.15% to the network and the remaining 0.95% split between the acquirer and the processor. The processor’s actual margin on that transaction might be as low as 0.3%.

This is why payment processors do not like to compete on price. The majority of the fee they charge is not theirs to cut. The only lever they control is their margin, which is already the smallest component.

What it means

Payment pricing is how five components, each controlled by different actors with distinct incentives, stack on top of one another.

This has three consequences:

First, price comparison in payments is fundamentally broken. Customers compare the one component they can see (the stated fee) and ignore the ones they cannot.

Second, payment companies that lead with transparency create a competitive advantage. When the entire structure is meant to obfuscate, customers are delighted by transparency. Wise has built a multi-billion-dollar business in large part by making the 2^nd Component (the FX markup) visible. This single design decision has primed the customers to expect transparency and forced competitors to respond. The next company to do this for another component (e.g., float or corridor cost) will likely have a similar advantage.

Third, if you are running a payments company and do not fully understand all five components of your own pricing architecture - how each component contributes to revenue, how customers perceive each component, and where the gaps between cost and price are widest - you are making pricing decisions on incomplete information. Your competitors who understand this will eventually take your customers.