How many PSPs do you need? Choosing providers per market

“How many PSPs should we have?” is usually asked as a number. The honest answer is that the number falls out of a different question: what jobs do payments have to do in this market, and which of them can you afford to lose for a day? A merchant that answers that second question per market ends up with a provider list it can defend, instead of one that grew by accident or one that stopped at a single contract.

This article sets out a method, shows how cost and redundancy pull in opposite directions, and works through an example for Mexico, with a shorter note on Uzbekistan. The provider counts in the examples are illustrations, not recommendations for any specific provider.

Why is “how many PSPs” the wrong first question?

Because providers are not interchangeable units. One provider may acquire cards domestically in Mexico but offer no cash vouchers. Another may be strong in bank transfers and pay out to them, but acquire cards cross-border. A third may accept your merchant category in one country and not the next. Counting providers before you know what you need them for tends to produce either gaps (no refunds or payouts for the method customers pay with) or overlap (three providers doing the same card job).

The better unit is the job: a specific thing a provider does in a specific market, such as “acquire Mexican Visa and Mastercard cards domestically” or “pay out to Mexican bank accounts”.

What jobs does each market need?

Most markets reduce to five jobs:

1. Card acquiring, ideally domestic. Stripe’s guide on authorization rates notes that banks are often more likely to approve domestic payments, which is why local acquiring can raise acceptance, and that local accounts can remove cross-border fees. If your current provider acquires cross-border in a market where a competitor acquires locally, that is a job your setup is doing badly.
2. Each local method with real volume. The local payment methods customers actually use, from instant bank transfers to wallets and cash vouchers.
3. Payouts. Refunds to methods that do not support them natively, marketplace seller payouts, or cashback. Not every pay-in provider pays out, and not every method can receive money.
4. Category acceptance. Providers accept or restrict merchant categories method by method. Stripe’s OXXO documentation, for example, lists merchant category codes that are prohibited or restricted for that method. In regulated and high-risk sectors, every provider for every job also has to accept your category and licence.
5. Continuity. A second route for any job whose failure would stop meaningful revenue.

The minimum number of providers is the smallest set that covers jobs 1 to 4. Job 5 decides how many you add on top.

When does a second provider for the same job pay off?

When the expected cost of losing that job exceeds the cost of running a second provider. A simple way to frame it per job:

Value of a backup ≈ revenue through the job per day × days of disruption you expect in a year × share you would lose

Then compare it with the yearly cost of the second provider: fees above what the primary would charge, any minimums, integration and reconciliation work, and the effect of splitting volume on pricing tiers. Add the softer benefits: a second provider makes approval rates comparable on the same traffic, keeps the first provider’s pricing honest, and gives you somewhere to go if an account is reviewed or closed. Our article on single-PSP risk covers those in more detail.

Two rules of thumb follow:

• The job that carries most of the volume almost always deserves a backup. In most markets that is cards or the dominant local method.
• A small method rarely deserves two providers. If a method carries a few percent of payments, the customer who cannot use it today can usually use another method in the same checkout.

What do more providers cost?

More than the headline fees. Each additional provider adds:

• A contract and due diligence, which in high-risk categories can take months.
• Pricing effects. Volume split across providers may land in worse tiers at each, and some providers set monthly minimums.
• Reconciliation. Another settlement file, fee structure, currency schedule and possibly another reserve.
• Monitoring. Another set of approval rates, error rates and chargeback ratios to watch, on smaller samples that move more from month to month.
• Integration, unless the providers sit behind one orchestration layer, in which case adding one is configuration rather than a project.

None of these is a reason to stay with one provider. They are the reason to add providers per job, not per instinct.

Worked example: how many providers for Mexico?

Example figures and provider names, for illustration only.

An online retailer that also runs a marketplace for third-party sellers is entering Mexico. Mexico pays on three rails at once, which our Mexico guide describes in more detail:

• SPEI, Banco de México’s real-time transfer system. Expansión reports that it processed more than 7.3 billion transfers in 2025 and that about 73.5 million adults use it. SPEI supports pay-ins and payouts.
• Cards, where domestic acquiring matters for the reasons above.
• Cash, still the main payment method for 85% of people on purchases of 500 pesos or less, according to the national financial inclusion survey cited by Expansión. OXXO Pay turns an online order into a voucher paid at a convenience store. Stripe’s documentation notes that vouchers expire after 5 days by default, payment is confirmed by the next business day, amounts run from 10 to 10,000 MXN, and OXXO payments cannot be refunded.

The merchant expects this mix of payments: SPEI 50%, cards 35%, OXXO 15%. Seller payouts go to bank accounts by SPEI, and so do refunds for OXXO orders, because OXXO payments cannot be refunded through OXXO.

Job	Volume share	Candidate providers	Decision
SPEI pay-ins	50%	A, B	Two providers; it is the largest job
Domestic card acquiring	35%	A (domestic), C (cross-border)	A as primary; C as backup for technical failures only
OXXO pay-ins	15%	B	One provider; customers can switch to SPEI if it fails
SPEI payouts	Seller payouts and OXXO refunds	A, B	Both; sellers and customers cannot wait for one provider to recover

The answer for Mexico is three providers, with A and B each covering two jobs. Now check the backup arithmetic for SPEI pay-ins. If the merchant takes MXN 2,000,000 a day in payments, SPEI carries MXN 1,000,000 a day. Two days of provider disruption a year with half of that volume lost costs about MXN 1,000,000 a year. If the second SPEI provider costs well under that in extra fees and work, it pays for itself, before counting the approval-rate comparisons and negotiating leverage it brings.

Run the same test on OXXO and the answer changes: MXN 300,000 a day, a customer base that can switch to SPEI in the same checkout, and a method that confirms payment by the next business day anyway. A second OXXO provider is hard to justify at this volume.

How does the answer change in CIS markets?

The jobs are the same; the rails are different. In Uzbekistan, for example, our Uzbekistan guide describes most demand running through the domestic card schemes Humo and Uzcard, often inside wallet apps such as Payme and Click. The Central Bank reported more than 68.27 million bank cards in circulation on 1 January 2026, according to Kun.uz. There, the core job is domestic card acceptance through local payment organisations, international cards are a smaller job, and payouts to local cards are their own job. A merchant might reasonably end up with two local providers for the domestic card job and one international acquirer, because the domestic job carries most of the volume.

How do you keep a multi-provider setup manageable?

• Write eligibility rules per provider: country, method, currency, amount limits, customer group and customer verification level, so a payment never reaches a provider that cannot take it.
• Route by segment, not by habit, using approval rates by card country, method and customer type.
• Keep every provider warm with a share of real traffic.
• Normalise statuses and webhooks so finance sees one transaction model, whichever provider processed the payment.
• Review the list every quarter. Drop providers whose jobs have moved elsewhere, and add a backup where a job has grown.

How does Omno help?

Omno puts one API in front of the providers for each market, for Pay-In and Pay-Out. Eligibility rules by country, method, customer group and customer verification level decide which providers may take each payment, and routing picks among them. If a provider is unavailable or rejects the request, the payment moves to the next provider in the rule until one accepts. The dashboard shows approval rates per provider, method and country on the same definitions. Adding or switching a provider is configuration rather than a new integration. The payment orchestration page shows how providers are configured.