Which local payment methods to add first: a scoring model

A merchant expanding into five new markets can make a list of twenty local payment methods in an afternoon. Integrating them is another matter. Each method means a provider contract, testing of payments, refunds and webhooks, a new settlement report and a support script. The team can add one or two a quarter, so the real question is which ones first.

This article sets out a scoring model for that decision: the gates to check before scoring, eight criteria with weights, sourced data points for reach in several regions, and a worked example across one market set. How methods should be ordered on the payment page once they are live is a different question, covered in checkout localisation; how many providers each market needs is covered in how many PSPs per market.

Why use a scoring model at all?

Because the loudest request usually wins otherwise. A country manager asks for one method, a large customer asks for another, and a provider offers a third as part of a bundle. A score makes the trade-offs explicit: a method used by most local shoppers but expensive to integrate, against one that is cheap to add but serves a small market.

The model has three steps:

1. Gates. Remove candidates that no provider can offer to your business today.
2. Method score. Rate each remaining method from 1 to 5 on eight weighted criteria.
3. Market weight. Multiply by the market’s share of your planned revenue, then integrate in that order.

What should you check before scoring?

Three gates, each a yes or no. A method that fails one is not ready to be scored.

• A provider can serve your legal entity. Many methods are reachable only through providers that support your account country. Stripe’s Pix documentation, for example, lists the business locations that can accept Pix from Brazilian customers and the settlement currency for each account country, such as USD for the US and EUR for most of the EU.
• Your business category is allowed. Providers keep their own lists of prohibited and restricted merchant category codes per method; Stripe publishes one for Pix. Check the method’s list against your MCC before scoring it.
• The setup you need exists. Some methods need a local contract. Checkout.com’s documentation says that to process mada cards within Saudi Arabia, a merchant must first obtain a merchant ID from a Saudi card acquirer. Outside Saudi Arabia, mada cards behave as debit cards on Visa or Mastercard.

Gates also catch scheme changes. Stripe tells merchants using iDEAL to rebrand their integration to iDEAL | Wero in the first quarter of 2026 and to switch to Wero during 2026 and 2027. Build for where the method is going, not where it was.

How do you measure how many customers use a method?

Start from public statistics, then narrow to your own customers. A method’s national reach is a ceiling; your category, ticket size and customer age decide how much of it you see.

Market	Method	Data point	Source
Brazil	Pix	42% of e-commerce value in 2025, against 40% for credit cards	Poder360, citing the Worldpay Global Payments Report 2026
Saudi Arabia	mada	SR101.4 billion of e-commerce spending in Q2 2026, up 32.3% on a year earlier	SAMA, via Saudi Gazette
Kazakhstan	Kaspi	28.4 million of 84.0 million payment cards on local systems such as Kaspi.kz, 1 August 2026	National Bank of Kazakhstan
Kenya	M-Pesa	95.31 million registered mobile money accounts across providers, August 2026	Central Bank of Kenya
Poland	BLIK	No recent official share found; Stripe describes it as a common payment method in Poland	Stripe

Not every data point is a share. The mada figure is spending, the Kenyan figure counts accounts rather than people, and for BLIK we found no recent official share. Score reach from 1 to 5 on the best evidence you have, note the source next to the score, and confirm it with your own data: payment step drop-offs, support requests for a method, and the provider’s figures for your category. Our articles on Pix and Kaspi go deeper on two of these markets.

Which criteria go into the score?

Eight criteria, each rated from 1 (poor) to 5 (strong), with weights that add up to 100. The weights below are a starting point; change them to match your business.

Criterion	What it measures	What to ask or check	Weight
Reach	Share of target customers who use the method	Central bank and operator data, your checkout data	25
Conversion uplift	Orders gained, not just shifted from cards	Drop-offs at payment, provider benchmarks for your category	15
Ticket-size fit	Whether limits and habits suit your basket	Per-payment and per-customer limits	10
Cost	Fees, FX, cross-border surcharges, settlement time	Full price list, settlement days, currency	15
Refunds and disputes	Refund windows, partial refunds, dispute process	Provider documentation per method	10
Payout support	Whether you can pay customers or sellers on the same rail	Payout availability and limits	5
Integration and operations	Build, testing, async statuses, reconciliation	Redirects, timeouts, webhook model, report format	10
Provider availability	Number of providers that can offer it to you	At least two options, or one you trust	10

Raise the payout weight for marketplaces that pay sellers, and the refund weight for categories with high return rates, such as fashion.

How do cost, refunds and disputes differ between methods?

Widely, even between methods that look alike at checkout. The table uses Stripe as one reference provider; other providers set their own terms.

Method	Fee on Stripe’s US price list	Settlement	Refunds	Disputes
Pix	2%	Up to 2 business days (T+2)	Full and partial, within 90 days	Fraud claims through MED; cannot be contested
BLIK	1.9% plus USD 0.30	Standard payout timing	Full and partial	Yes, with 12 calendar days to submit evidence
iDEAL	USD 0.80	Standard payout timing	Full and partial, up to 180 days	Customers cannot dispute with their bank

Stripe adds 1.5% for international transactions and 1% when currency conversion is needed. Pix also carries a decision on tax: Stripe says Brazilian customers paying businesses outside Brazil pay 3.5% IOF unless the merchant absorbs it, and caps cross-border Pix at 3,000 USD per day and 10,000 USD per month per customer per business.

Flow details belong in the effort score. Stripe says a BLIK code is valid for 2 minutes and the customer has 60 seconds to approve in the banking app, while Pix has no redirect back from the banking app, so the checkout must wait for a final status. For refund rules on more methods, see refunds across local methods.

Worked example: scoring five methods for one expansion plan

Example figures, for illustration only.

A European online fashion retailer plans next year’s revenue from five new markets: Brazil 35%, Saudi Arabia 25%, Poland 20%, Kazakhstan 10% and Kenya 10%. It already takes cards everywhere and can add two methods this half-year. It scores one candidate per market, using the weights above.

Criterion (weight)	Pix	mada	BLIK	Kaspi	M-Pesa
Reach (25)	5	5	4	4	5
Conversion uplift (15)	5	4	3	4	4
Ticket-size fit (10)	4	5	4	4	3
Cost (15)	3	3	4	3	3
Refunds and disputes (10)	4	3	4	4	3
Payout support (5)	4	2	2	2	5
Integration and operations (10)	4	2	4	3	3
Provider availability (10)	5	3	5	2	3
Method score (out of 5)	4.35	3.70	3.85	3.45	3.75

Each method score is the sum of weight times score, divided by 100. For Pix: 125 + 75 + 40 + 45 + 40 + 20 + 40 + 50 = 435, so 4.35.

Method	Method score	Market share of planned revenue	Priority (score × share)	Order
Pix	4.35	35%	152.3	1
mada	3.70	25%	92.5	2
BLIK	3.85	20%	77.0	3
M-Pesa	3.75	10%	37.5	4
Kaspi	3.45	10%	34.5	5

On method score alone, BLIK would come second, ahead of mada. Weighted by revenue, mada moves up because Saudi Arabia carries more of the plan, even though it needs a local acquiring setup. The retailer adds Pix and mada first, measures uplift for a quarter, then re-scores BLIK, M-Pesa and Kaspi with its own data. If Kenya’s share of the plan doubled to 20%, M-Pesa’s priority would rise to 75.0, close to BLIK’s.

How do you turn the scores into a roadmap?

Treat the ranking as a queue, not a verdict.

1. Integrate the top one or two. Test payments, refunds, async statuses and reconciliation before launch.
2. Measure on payments, not attempts. Compare conversion at the payment step and total approved orders for the market before and after. Our guide on how to calculate approval rate explains the difference.
3. Replace estimated scores with measured ones. Uplift and cost become real numbers after a quarter.
4. Re-score the queue. Fees, providers and plans change, and a method that ranked fifth may move up.
5. Add a second provider where it pays. Once a method carries real volume, provider availability becomes a resilience question, covered in single PSP risk.

What are the common mistakes?

• Scoring before checking the gates. A perfect score means nothing if no provider will onboard your entity or category for that method.
• Using national reach as your reach. Central bank figures describe the whole market. Your customers may be younger, richer or more card-heavy.
• Counting shifted orders as uplift. A shopper who would have paid by card adds cost, not revenue, when they switch to the new method.
• Ignoring the cost of money arriving later. Settlement time and FX conversion belong in the cost score, next to the fee.
• Forgetting refunds and payouts. A method that cannot refund within your returns window will cost support time later.
• Never re-scoring. Schemes change: iDEAL’s move to Wero is one example of a method changing under live integrations.

How does Omno help decide which local methods to add first?

Omno can put several providers and their local methods behind one API, so adding a method from a provider that is already connected is configuration rather than a new integration, with routing rules by country, method, currency and customer group. The dashboard shows volumes, approval rates and decline reasons by provider, method and country, and data export lets you measure uplift after launch and feed the next scoring round. The merchant still negotiates provider contracts and fees. The payment orchestration page covers the setup.