At the end of every month, someone in your finance team opens a spreadsheet. They export a settlement report from one PSP, download a CSV from another, pull a third file from a regional acquirer, and spend hours trying to make them agree. Transaction IDs do not match. Fee labels are different. Timestamps are in different time zones. The number that lands in the general ledger is, at best, an approximation of what actually happened.
Most enterprise merchants treat reconciliation as an operational task to be automated. But the expensive reconciliation problems are not the ones that get caught. They are the ones that do not.
Reconciliation rarely shows up on a list of major costs, because its cost is almost never captured as a line item. It gets absorbed into headcount, into month-end hours, into the assumption that this is simply what finance work looks like.
The numbers tell a different story. Modern Treasury's 2025 State of Payment Operations report found that 98% of companies still perform some payment operations manually, and 51% perform up to half manually. 49% use five or more systems to manage payments, and 68% of finance teams say they waste significant time on payment operations.
The 2025 AFP Treasury Benchmarking Survey, underwritten by Wells Fargo, reports that nearly three-quarters of treasury practitioners cite cash management and forecasting as their top priorities. Reconciliation is often exactly what prevents them from getting there. Every hour spent normalizing data from five PSP formats is an hour not spent on cash flow analysis or cost optimization.
Acquiring fees vary by card brand, transaction type, and geography. When a transaction is misclassified, the difference can be 20 to 40 basis points, and it disappears inside the aggregated statement.
According to Optimus Fintech, merchants who validate fees at the transaction level routinely identify overcharges of 0.1% to 0.3% of processed volume: between $200,000 and $600,000 annually on $200 million in throughput. For context, U.S. merchants paid $187.2 billion in card processing fees in 2024.
Cardholder dispute windows run up to 120 days, and merchants have as little as 30 days to respond, according to Visa and Mastercard dispute management guidelines.
When transaction context lives in five systems and the response process is manual, assembling a case file can take longer than the window allows. The merchant does not lose the dispute on the evidence. They lose it on the clock.
A refund processed twice because two reps handled the same ticket. A capture that settled at a different amount than the original authorization. Each one is small. Each one compounds.
Each PSP exposes payment events through its own schema. A single transaction can generate three or four distinct records: one when the payment intent is created, another when it is captured, another when the fee is calculated, another when the payout settles. Each record carries a different identifier, and the relationships between them are not one to one. A capture can split across multiple settlements.
All of these events have to be mapped to a single internal model before any matching can happen. According to Kani Payments' 2025 Reconciliation and Reporting Survey, the average firm spends three hours just preparing data before reconciliation begins. Cross-currency matching was cited by 23% of respondents as their top challenge, and 82% struggle to meet reporting timelines.
As providers are added, the reconciliation burden grows while the team handling it stays the same size.
Transaction records, settlement reports, fee breakdowns, refund histories, and chargeback statuses live in different systems, in different formats, on different timelines. Connecting them after the fact is what creates the monthly reconciliation burden. Connecting them at the source is what eliminates it.
An orchestration layer that sits above multiple payment providers and normalizes transaction data as it flows through makes matching happen while transactions move, not weeks later. That single change turns all four problems above from discoveries into corrections: fee discrepancies surface against contracted terms, dispute evidence is assembled while the window is still open, duplicate refunds are caught at the second attempt, and data preparation disappears because the data was never fragmented to begin with.
It also changes what the finance team spends its time on. Cash visibility becomes real rather than approximate, which is the foundation for the forecasting work that treasury teams say they want to prioritize and rarely reach.
This is what DEUNA's reconciliation capability is designed to enable. It sits inside the same orchestration layer that handles routing, so the data driving payment decisions is the same data reconciliation runs on. On top of that, Athia, DEUNA's agentic payments intelligence layer, turns that data into action: spotting opportunities as they happen and showing where providers are quietly costing you money.
Fixing reconciliation as a process buys you a few hours back every month. Fixing it as an architecture buys you the hours, the leakage, and a set of numbers your finance team can actually act on.
