The customer renewed. The invoice went out. The payment failed. For a B2B SaaS business entering new markets, that gap can turn an encouraging sales forecast into an uncomfortable cash meeting. The immediate question is whether the money will arrive late, disappear with a cancelled subscription, or prove to be a fraudulent transaction that should never have been accepted.
The Forecast Risk: International expansion changes the mix of issuing banks, currencies and payment routes behind subscription collections. Changes to fraud controls can also alter which transactions reach approval. Finance teams should separate temporary collection failures from permanent payment problems and suspected fraud. Forecast recoveries by customer cohort and collection date, then deduct processing costs and expected losses. A higher approval rate earns its place in the budget only when it produces additional, profitable collections.
Cross-Border Growth Changes the Assumptions Behind Collections
A domestic approval rate is a weak starting point for an international forecast unless the business tests whether it transfers. New customers may bring different card issuers, corporate spending controls and billing currencies. The payment route may also change. Those differences give finance a reason to build a separate collection assumption for each new market.
Consider a hypothetical software company expanding from the UK into continental Europe and North America. Its commercial forecast projects subscription demand. Its collection forecast needs another layer: how much billed value will clear through the payment arrangements available in each market, and when?
The distinction matters most for businesses collecting subscriptions by card. Enterprise contracts paid by invoice and bank transfer need separate assumptions. Combining both groups into one payment success figure can hide a problem affecting smaller, card-paying customers.
There is also a measurable fraud difference across payment geographies. In their December 2025 payment fraud analysis, the European Banking Authority and European Central Bank reported that card fraud rates were about 17 times higher when the counterpart was outside the European Economic Area than for domestic transactions. The report covers data from 2022 to 2024 and takes an EU/EEA card-issuing perspective. It also found generally lower fraud rates for card payments using strong customer authentication.
Those findings are context, not a SaaS approval benchmark. They do not establish how much any particular expansion will reduce acceptance. The business implication is narrower: changing geography deserves a fresh assessment of both collection performance and fraud exposure.
A Decline Needs a Diagnosis Before It Enters the Forecast
A failed payment is an outcome, not a complete explanation. Finance should resist placing every unpaid renewal into either guaranteed recovery or immediate churn. Both choices can distort the outlook.
A useful working classification separates the action required from the eventual commercial result:
- Potentially recoverable failures: Temporary funding or technical problems may clear after an appropriate delay. Recovery still needs evidence from comparable accounts.
- Customer-action failures: An expired credential, spending restriction or authentication requirement may need the customer's involvement before collection can succeed.
- Permanent credential problems: A closed account or unusable card calls for a replacement payment method rather than repeated submissions.
- Suspected or confirmed fraud: Suspicious activity requires investigation. Confirmed misuse belongs in the loss and prevention analysis, not the recovery target.
A legitimate customer blocked by a fraud rule is a false positive. A criminal using stolen payment details is genuine fraud. A vague issuer response may establish neither. Treating an ambiguous decline code as proof of fraud can understate recoverable business. Treating it as permission to retry indefinitely can overstate collections and create unnecessary exposure.
For B2B subscriptions, the recovery process should identify who can fix the problem. The product user may have no authority over the company card. A clear billing message to the account's payment administrator can therefore be a more useful intervention than another automatic attempt.
Fraud Rule Changes Can Move the Forecast Without Changing Demand
Suppose a payments team tightens a fraud threshold during an international launch. More legitimate transactions could be blocked alongside suspicious ones. Finance might interpret the resulting collection shortfall as weaker customer demand, especially if the rule change is absent from management reporting.
The reverse deserves equal scrutiny. Relaxing a rule could lift approvals while accepting transactions that later become losses. An approved payment alone cannot establish that the customer is legitimate or that the collection will remain profitable.
The regulatory findings on authentication support keeping fraud protection in the calculation. They do not establish the right threshold for an individual SaaS company. That decision needs the company's own results, separated by market and customer type.
A sensible reporting practice is to date every material fraud-policy change. Track the affected cohort's approvals, completed collections and later disputes against a comparable baseline. Keep early results provisional until enough time has passed to assess losses. Otherwise, the budget may recognise the apparent benefit before the associated cost becomes visible.
Routing and Retries Need a Commercial Hurdle
Midway between accepting every decline and chasing every failed payment sits a more selective approach. Payment teams assessing approval rate optimisation should examine whether better transaction data, an alternative acquiring route or a properly timed retry addresses a specific failure. The supplier's undated feature description discusses segmented baselines, routing and controlled experiments. Its illustrative performance ranges should not become a forecast assumption without merchant-specific validation.
Routing merits a test when the existing arrangement appears to perform poorly for a defined issuer, currency or card segment. Another acquirer may offer a better result. That possibility needs to be weighed against integration work, commercial terms, reconciliation effort and ongoing maintenance.
Retries need their own logic. A temporary problem may justify a later attempt where the processor's instructions permit it. A request for authentication calls for the correct authentication flow. A permanently unusable credential calls for updated details. Confirmed fraud calls for prevention and investigation.
Alternative routes must preserve the transaction's true context and required controls. Moving a suspicious payment between acquirers to obtain approval would undermine the purpose of the test. The commercial objective should be additional legitimate collections after costs and losses.
Implementation spending also belongs in the decision. Engineering time has a cost even when no new supplier fee appears on the invoice. Finance and payments teams should agree the minimum gain needed to repay that work before expanding a pilot.
Separate Cash Timing From Subscription Revenue
Consider a hypothetical monthly cohort with £1 million of card-billed renewals. Assume 94% clears initially. That leaves £60,000 unpaid at the first attempt.
If half of the failed value is recovered within seven days, eventual collections reach £970,000. The initial £60,000 shortfall was neither entirely lost nor entirely harmless. Some cash arrived late. Another £30,000 remained outstanding. A month-end cut-off could place those amounts into different reporting periods.
Now assume a new international cohort achieves only 90% initial collection on the same billed value. If half of the failed value is again recovered, collections reach £950,000. Applying the original cohort's eventual 97% collection assumption would overstate expected cash by £20,000. These figures are illustrative, not industry averages.
The model should distinguish billed value, collected cash and recognised revenue. A payment decline does not, by itself, determine the accounting treatment of a subscription. Finance needs to apply the relevant contract terms and accounting policy. Operational reporting should also separate an account awaiting payment from a confirmed cancellation.
Recovery timing deserves an explicit assumption. A payment expected next month cannot fund this month's obligations. An annual subscription collected late can create a particularly visible cash variance, even where the customer continues using the service.
Measure Collection Success Without Letting Retries Confuse It
Approval rates need clear denominators. An attempt-based measure divides approved authorisations by all authorisation attempts. An invoice-based collection measure asks how many invoices, or how much billed value, have been paid within a defined period.
Imagine 100 invoices produce 90 approvals initially. Ten retries then recover five more invoices. Attempt-based approval is about 86%, calculated as 95 approvals across 110 attempts. Invoice collection success is 95%. Both figures describe the same hypothetical activity, but answer different questions.
For forecasting, track first-attempt performance alongside collection within fixed windows, for example seven and 30 days. Also measure value, because collecting several small invoices may not compensate for one failed large renewal.
Keep the customer mix visible. A growing share of new international accounts can move the blended approval rate even when each established cohort performs consistently. Without that breakdown, management may attribute a change to routing or fraud policy that actually reflects expansion.
Build the Forecast Around Recoveries and Their Full Cost
A forecast can handle payment uncertainty without pretending every decline has a known outcome. The aim is to connect each assumption to an observable cohort and update it as evidence arrives.
- Establish a dated baseline. Record billed value, initial collection and subsequent recovery by market, payment method and customer cohort.
- Model separate outcomes. Use base, downside and upside recovery assumptions, with explicit collection dates and unresolved amounts.
- Test a defined intervention. Compare a routing, retry or fraud-control change against similar traffic while keeping other material settings stable where possible.
- Reconcile the commercial result. Deduct incremental processing, implementation, support and fraud costs before transferring a measured gain into the forecast.
Processing contracts need careful reading. Depending on the agreement, costs may attach to attempts, successful transactions, cross-border activity, currency conversion or disputes. A business should model its actual terms rather than assume every failed attempt is free.
For example, an illustrative pilot that collects an additional £20,000 has not automatically produced £20,000 of benefit. Additional fees, losses, service-delivery costs and the allocated implementation expense reduce its contribution. Conversely, fewer billing support cases could add value if that saving is measured.
Make International Growth a Collectable Number
The forecast becomes more credible when sales demand, payment performance and fraud exposure meet in the same model. Each new market needs a collection assumption. Each material control change needs a date. Each recovery target needs evidence about which failures can be resolved and how long that takes.
For executives, the decisive question is straightforward: how much legitimate customer value will become cash, by what date, and at what cost? Answering it keeps recoverable declines in view without treating genuine fraud as missed revenue. It also gives international expansion a firmer financial basis than a blended approval rate can provide.
