United KingdomPublic sector2026 – ongoingHM Revenue & Customs
Business Propensity to Pay debt segmentation
HMRC scored approximately 4 million business debtors monthly to tailor collection communications
Problem
Most business debtors previously received similar collection communications. An interim segmentation model focused on pandemic impacts rather than repayment behaviour.
Approach
HMRC linked business tax debts and trained gradient-boosted decision trees on historical repayment behaviour. Monthly scores fed debtor segments into collection systems, helping determine the timing and content of letters. Staff could view scores when reviewing cases, while aggregate monitoring checked changes in model outputs.
Results
Takeaway
HMRC used repayment predictions to tailor an existing collection process. The disclosed scale did not establish an increase in tax recovered.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- Finance, risk and legal
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
HMRC reported monthly scoring of approximately 4 million customers. Predictions informed collection communications alongside other information; no recovery uplift was quantified.
Full details are in the original sources.