Card issuers measure fraud losses closely. The cost of the legitimate transactions their fraud systems wrongly decline is another matter. Analysis published by payment technology provider BPC on 6 August 2026 puts that cost at an average of $160,000 a year in lost interchange revenue for a mid-sized issuer, modelled on 10 million debit transactions a month and a deliberately conservative false decline rate of 0.5 percentage points.

Khurram Ahmed, senior product consultant, fraud solutions at BPC, answered written questions from The Fintech Times on why that cost goes unmeasured, what a realistic false decline rate looks like for a typical issuer, and what a migration from rules-based authorisation to real-time decisioning involves in practice.

Ahmed’s explanation for the measurement gap is structural. “Banks have spent decades building frameworks to measure fraud. Every fraudulent transaction is tracked, reported. False declines do not produce the same visibility.” A wrongly blocked legitimate transaction typically shows up as a customer service issue or a temporary inconvenience, not as a line in the finance report. The cost sits across several areas at once: lost interchange, customer friction that pushes spend to other cards, and the operational overhead of handling disputes. Most banks never connect those fragments into a single figure.

For false declines to become a board metric, he says, banks need to measure approval accuracy alongside fraud detection: tracking false decline rates by product, channel and customer segment, then calculating the revenue impact of the blocked legitimate spend. “Once the data is structured and visible, it becomes possible to set targets and hold teams accountable.”

How wide the spread is

The 0.5 percentage point rate in BPC’s modelling was chosen to show the minimum exposure. Ahmed’s view of a typical mid-sized issuer on a legacy rules-based system is considerably worse: a ratio of around 30 to one, meaning roughly 30 good transactions wrongly declined for every fraudulent one caught. The spread between the best and worst performers is wide. Banks running modern real-time fraud platforms with continuous profiling and adaptive models can hold false declines below 0.5 per cent, while issuers relying on static rules and limited transaction context can see rates above 5 per cent, particularly during peak periods or when customers travel. “The difference comes down to platform capability and operational discipline.”

BPC’s analysis picked out South Africa and Ecuador as the markets where issuers lose most, and Ahmed frames those as representative rather than exhaustive. “It happens everywhere.” Higher-interchange markets lose more because the revenue attached to each transaction is larger: he puts credit card interchange at up to 1.48 per cent in South Africa and around 1.36 per cent in Ecuador, against European caps of 0.30 per cent for credit and 0.20 per cent for debit. That changes the modernisation case. “For issuers in high-interchange markets, the revenue leakage from false declines is a direct hit to the business model.” Every percentage point of improvement in approval accuracy translates into measurable income, which makes the return on a fraud platform upgrade faster and the cost of delay higher.

Where the rules go wrong

Legacy rules-based systems apply fixed thresholds to transaction attributes. Spending above a set amount, a transaction in a new country or several purchases inside a short window trigger a decline or a manual review. The problem, Ahmed says, is that “these rules cannot account for context”. A customer who normally spends £50 a transaction and suddenly spends £500 might be buying a flight or replacing a broken appliance, and can be authenticated in real time through a challenge rather than declined outright. Rules also decay: spending patterns that were unusual five years ago are now routine. The typical failure points are travel, large purchases, first-time online transactions and rapid repeat spending, all scenarios where legitimate behaviour looks like fraud if the system lacks the context to tell them apart.

Migration without switching everything off

The false declines analysis was published in BPC’s guide Modernisation Without Disruption, and Ahmed’s account of migration follows its argument. A realistic move to real-time decisioning does not require an issuer to replace its whole legacy stack at once. For many mid-sized issuers the lower-risk approach is phased: map the existing transaction flows, BINs, data and integrations, run mock migrations and testing, then move selected products, customer groups or components in controlled waves, with legacy and new platforms running in parallel for a period and reconciliation and rollback criteria in place.

For fraud management specifically, a real-time decisioning layer can be introduced alongside existing systems before the full issuing environment is replaced, with BPC’s SmartVista platform acting as an intermediary between legacy and modern components. “For the customer, cards, wallets, authorisations and settlement continue to work while the technology underneath is modernised.” BPC says it has applied the approach across more than 400 legacy migrations in more than 100 countries.

Regulators and false declines

Asked whether regulators pushing for tougher fraud controls are in tension with issuers trying to cut false declines, Ahmed accepts there is tension but not a fundamental conflict. Regulators want fraud prevented and customers protected; banks want legitimate transactions approved with less friction. Both are achievable if the fraud platform is accurate enough in real time. The problem arises when banks reach for blunt instruments to satisfy a requirement. “Tightening rules to reduce fraud often increases false declines unless the underlying decisioning logic improves.” His prescription is to treat fraud prevention “as a precision problem, not a volume problem”, assessing risk at the transaction level with behavioural data, device intelligence and contextual signals, and to report fraud loss and false decline rates together as joint performance indicators.

Datos Insights estimates that false declines cost the industry $213 billion in 2025 and could reach $297 billion by 2029. Asked which single change would bend that curve fastest, Ahmed chooses the move from static fraud rules to real-time, adaptive, customer-level decisioning, which he notes is also the direction Datos Insights points to. Banks that have made that migration, he says, “consistently report false decline reductions of 20 per cent to 40 per cent without increasing fraud losses.” The obstacle is that many issuers are still running fraud infrastructure built 10 or 15 years ago, and modernising it requires investment and operational change. “Until that happens, the false decline cost will continue growing.”

BPC’s Modernisation Without Disruption guide, which sets out four migration strategies and the case studies Ahmed refers to, was published alongside the false declines analysis on 6 August 2026 and is available on the company’s website.

The post BPC: Banks Count Fraud Losses But Not What False Declines Cost appeared first on The Fintech Times.