The Team Thought They Had a Model Problem. They Had a Compliance Problem.
The model wasn't broken—the foundation was. Fixing compliance in time for Black Friday lifted approvals from 54% to 72%, driving ~100M DKK (Danish Krone) in GMV (gross merchandise value).
Most BNPL credit teams optimize in the wrong order. They tune models before they can trust their data. They chase approval rates before they've satisfied regulators.
They build on a foundation that hasn't been stress-tested — and then wonder why nothing moves.
54%
Approval Rate
Oversimplified view
The company came in with a clear complaint: approval rates were stuck at approximately 54%, and revenue was being left on the table. The internal team's diagnosis was straightforward — fix the credit model. That diagnosis was wrong.
A full audit revealed three compounding problems the team had overlooked:
No data traceability.
Decisions couldn't be reconstructed or audited. There was no record of why a loan was approved or declined — making the system indefensible under scrutiny.
No regulatory alignment.
The credit engine didn't meet FSA-level compliance standards for affordability assessment. In the Nordic regulatory environment, financial supervisory authorities hold BNPL providers to strict consumer protection requirements. This system didn't qualify.
No monitoring infrastructure.
No dashboards, no risk visibility, no way to observe system behavior in production. The team had no instrumentation to know what the system was actually doing.
Improving the model on top of this infrastructure wouldn't have raised approval rates. It would have amplified the risk of a regulatory incident while producing numbers no one could trust.
It all started with getting the diagnosis right:
Root cause analysis.
Identified that the team was solving a symptom — low approvals — rather than the cause: untraceable, non-compliant, unmonitored infrastructure.
Compliance and traceability first.
Reframed the entire project around regulatory alignment. Refactored the credit engine with full decision auditability. Every approval and decline could now be traced, justified, and reviewed by regulators.
Monitoring and risk dashboards.
Built real-time dashboards surfacing credit risk, portfolio performance, and approval patterns. For the first time, the team had operational visibility into what the system was doing and why.
Affordability model improvement.
With a compliant, traceable, observable foundation in place, the credit model was rebuilt using true affordability logic and statistical modeling — replacing heuristics with a data-driven, defensible approach.
Approval rate: 54% → 72% — an 18 percentage point increase
~100M DKK (Danish Krone) in GMV (gross merchandise value) processed during Black Friday 2025
Full FSA-level regulatory alignment achieved across the credit engine
Complete decision traceability implemented — every outcome auditable on demand
The approval rate increase was a byproduct of doing the foundational work correctly. Not of chasing the number directly.
Case Studies
This is just one example.
I have many more case studies with full technical walkthroughs, data, and methodology.
Work with me