Case Study: Building ML credit models and MVP lending pilots for Wave Money

Challenge

Wave Money, the leading payments app in Myanmar, aimed to launch lending products for customers without formal credit histories. Wave wanted to leverage customer transaction data to support lending decisions. However, lending was not part of the organisation's core roadmap, and resource constraints limited the ability to implement traditional infrastructure. Wave needed to move quickly with minimal investment, testing customer needs, product viability, and internal capacity through low-fidelity MVPs.

Approach

Factor provided end-to-end project management, guiding Wave through the design and implementation of lending pilots and the creation of a machine learning–driven credit scoring model, WaveScore. The approach prioritised manual workarounds and agile testing to accelerate learning while managing risk.

1. Lending Pilot Design and Execution

• Bank Pilot: Used to generate loan performance data and train the WaveScore model. The subsequent phase validated WaveScore as a strong predictor of repayment risk.

• NBFI Pilot: Focused on understanding customer needs. Two rounds tested different loan products

• “Negative Balance” Pilot: Aimed at enabling micro-credit to unbanked users via Wave account overdraft, with minimal manual handling.

2. Credit Score Development (Machine Learning Framework)

• Led the initial development of credit score using a machine learning model trained on loan repayment data.

• Designed data pipelines and features that fed into a machine learning framework that produced the credit score

• Recruited dedicated data and operational staff to support long-term credit score delivery and pilot scaling.

• Transitioned credit score ownership to Wave’s internal Data Science team, supported by external advisors.

3. Governance and Performance Review

• Established the WaveScore Steering Committee and Credit Review Forum to oversee strategy and score performance.

• Governance ensured WaveScore remained adaptive, data-informed, and aligned with evolving business needs.

4. MVP Operationalisation

• Designed manual workflows to launch lending pilots with limited tech integration.

• Coordinated pilot operations, including disbursements, repayment processes, collections, and reporting.

• Engaged cross-functional teams from Wave and its partners to implement, monitor, and adjust pilots.

Outcomes

• Built WaveScore as a machine learning credit score tailored for unbanked and underbanked customers.

• Validated its accuracy through live loan pilot performance and Credit Review Forum analysis.

• Created sustainable performance monitoring mechanisms that informed model updates and strategy shifts.

• Enabled Wave to test lending products at low cost with scalable learnings.

• Embedded a robust governance model for credit strategy and risk review.

Services Delivered

• Project governance setup and management, including steering committee leadership, project planning, reporting, and risk/issue tracking

• MVP lending pilot design and execution, across three pilot programs with tailored operational processes

• Machine learning–based credit score development, including feature engineering, model training, and pilot validation

• Loan performance analysis and governance structuring, including the WaveScore Credit Review Forum and decision pathways

• Cross-functional coordination, engaging Wave, 3rd party lending partners, and internal departments (legal, tech, ops, marketing)

• Resource planning and recruitment support, including onboarding of the Loan Ops Manager and an in-house data scientist

• Target operating model development, including assessment of “buy, build, partner” options, gap analysis, and resourcing implications

• Transition to internal ownership, with knowledge transfer and handover to Wave Data Science and operational teams