Risk is visible in different parts of the business.
Payment, billing, service and engagement data often sit in separate systems and are considered too late.
Energy debt
Act before arrears become entrenched.
Energy debt is not one problem with one cause. Earlier signals can help suppliers understand changing circumstances, identify risk and make appropriate support easier to reach.
The scale
Debt continues to rise.
The numbers underline the need for earlier, more individualised and more effective engagement.
£4.79bnDomestic debt and arrears in Q1 2026
+5%Increase from the previous quarter
+15%Increase compared with Q1 2025
£1,329Average balance, a record high
Sources: Ofgem debt and arrears indicators; BFY analysis of Ofgem’s Q1 2026 data. The analysis reports that £3.6bn, or 75%, had no repayment arrangement and that the average balance reached £1,329.
The problem
The signal often exists before the missed payment.
A balance is the outcome of a customer journey. Payment changes, bill shock, service issues, communication difficulties and changing household circumstances can all appear earlier.
Customers in debt have different circumstances.
Credit risk and vulnerability are distinct. Vulnerability must be considered separately and before any treatment decision.
The same intervention will not work for everyone.
Customers differ in their ability to pay, likelihood of repayment, communication needs and response to support.
Persistent arrears are harder to resolve.
Delay can increase customer harm, write-offs, complaints and the operational effort required to recover balances.
Responsible decisioning
Customer need must organise the response.
Commercial context can help a supplier understand the wider relationship, but it must not determine who receives appropriate affordability or vulnerability support.
Separate the decisions
Use debt-risk indicators to understand repayment risk. Use vulnerability information to identify support needs. Apply commercial measures only as additional context.
Measure the treatment
Understand whether contact or support genuinely improves the outcome, rather than assuming that more intervention always creates more value.
AI and analytics
Use sophistication where it improves the outcome.
The answer may be clear cohort analysis, segmentation, predictive modelling or uplift modelling. AI should earn its place by improving insight or action.
Debt-book and cohort analysis
Establish where debt is building, how customer groups differ and which journeys require closer attention.
Propensity and behavioural models
Surface customers more likely to experience payment difficulty or persistent arrears before the outcome is fixed.
Segmentation and uplift
Identify whose outcome is more likely to improve because of treatment, rather than those likely to improve without it or unlikely to benefit.
Monitoring and governance
Track model performance, treatment fairness and customer outcomes so decisions remain explainable and responsible.
Relevant proof
Debt insight designed for action.
Optima has helped energy and utilities providers modernise debt decisioning and move from broad treatment to more targeted, proportionate collections strategies.
Scalable debt propensity modelling
Optima migrated and productionised a business-critical debt propensity model in Databricks, embedding automated ingestion, testing and governance and integrating outputs into debt-management workflows.
- Faster, more automated recovery decision-making
- More reliable and scalable model operation
- Fairer and more consistent customer treatment
Customer debt segmentation
Optima developed behavioural and financial segments using payment history, arrears trajectory, days past due and customer engagement, then designed targeted strategies for each group.
- More targeted and proportionate collections
- Less unnecessary contact for customers likely to self-resolve
- Better support for customers experiencing financial difficulty
Start a conversation
Where is customer value leaking in your energy business?
We can help you understand the debt, churn, billing and customer-value signals already in your data, identify the highest-value opportunities and move from analysis to action quickly.