During a quarterly business review, the CFO requests the organization’s approval rate for the prior quarter. Payments presents one figure, fraud another, and finance a third. Each is methodologically sound. None of them align.
This is payment data fragmentation: data distributed across payment providers, fraud tools, checkout, and finance systems, each with its own formats and definitions. The issue is widespread. According to a January 2025 PYMNTS Intelligence report produced with Spreedly, 82% of online retailers report difficulty identifying the causes of failed payments.
Fragmented payment data rarely presents itself as a data problem. It surfaces as three warning signs in approval rate performance.
Without consistent definitions and a common time axis across providers, the same approval rate yields different results depending on which system calculates it.
The cost is time. When approval rates decline, teams spend the first days reconciling figures instead of diagnosing the cause. Meanwhile, issuer behavior may already have shifted: a bank tightens its fraud filters for a merchant category, and a BIN range that performed well the previous day begins to underperform.
Key question: Does every team report the same approval rate for the same period?
Hard declines, such as a closed account or a card reported as stolen, are final. Soft declines, such as a timeout or a temporary issuer restriction, can often be recovered through a retry or by routing to an alternative provider.
When each provider reports decline codes in its own format, recoverable declines are grouped with permanent ones, and recoverable revenue is written off. Retrying without that context is not a solution, as card networks restrict repeated retries and each attempt carries a cost.
The impact reaches the customer. Baymard Institute found that 10% of US online shoppers who abandoned a purchase did so because their card was declined.
Key question: What proportion of your declines are soft, and what share of those are recovered?
When providers report fees under different structures, finance cannot determine which one is most cost-effective by transaction type, market, or card. Processing costs rise without clear attribution.
Fraud intensifies the pressure. The LexisNexis Risk Solutions 2025 True Cost of Fraud study found that US merchants incur an average cost of $4.61 for every $1 of fraud. Fraud rules tighten in response, yet without unified data, there is no reliable way to measure how many legitimate customers those rules decline.
Key question: Can approval rate and cost per transaction be compared across providers on a consistent basis?
The solution sits beneath the reporting layer:
When a simple question about approval rate yields three answers, the issue is not insufficient data, but insufficient integration. For a business processing $1 billion annually, each percentage point of approval rate represents $10 million in revenue.
DEUNA connects enterprise merchants to more than 400 payment and fraud providers through a single integration and unifies payment data in real time, from checkout to settlement. Athia, DEUNA’s agentic payments intelligence, analyzes that data, identifies material changes in performance, and prioritizes the actions with the greatest impact.
