CaSE STUDY: A Leading Digital Bank in Africa.

A major African neobank serving millions of consumers with low-cost retail financial products designed for unbanked and underserved markets.

INDUSTRY
Neobanks/Challenger Banks
BUSINESS MODEL
100% Digital Bank in Africa
REGION
Africa
CUSTOMERS
Underserved, mass-market & underbanked populations
PRODUCTS
Personal loans, high-yield savings & consumer lending
CHANNEL
Expand credit access to thin-file customers while maintaining portfolio health

THE CHALLENGE

As the bank rapidly expanded its unsecured personal loan portfolio, it faced significant hurdles in maintaining portfolio health while driving financial inclusion

problem 1
High Risk Profile
Analysis of a shared sample revealed a 11.4% bad rate, indicating that existing scoring methods were not sufficiently weeding out high-risk applicants.
problem 2
Model Performance Ceiling
The institution's internal risk models reached a plateau, requiring a significant infusion of alternative data insights to improve predictive accuracy.
problem 3
Market Expansion Friction
Inaccurate scoring limited the bank's ability to safely expand credit access to thinner-file customers without compromising capital stability.

Why Incomplete Data Still Creates Growth Barriers for Lenders

To grow responsibly, lenders need a broader and more predictive view of applicant risk. Even with alternative data sources in play, many current models are still missing one critical layer: behavioural data that reveals true intent and repayment capacity.

Traditional Bureau Coverage
11%
Smartphone Penetration
50%
Credolab Scoreable Population
50%
Smartphone penetration vs. bureau coverage
Africa consumer base
Fintech in Africa: The end of the beginning McKinsey PDF

THE SOLUTION

Credolab Added a Behavioural Layer to the Existing Underwriting Stack.

STAGE 01
Acquisition
Mobile & kiosk
channels
STAGE 02
Onboarding
Identity, KYC, SDK init
STAGE 03 · CREDOLAB
Decisioning
Behavioural +
proprietary scorecard
STAGE 04
Disbursal
Loan booking & funding
STAGE 05
Servicing
Repayment & portfolio
monitoring

IMPLEMENTATION

Credolab was integrated into the client's decisioning workflow and delivered measurable
business impact within the first benchmarking cycle.

Historical data review

The client's existing portfolio and performance data were
analysed to establish a benchmark.

Signal generation

Credolab generated behavioural and device-based risk signals from applicant data.

Model validation

The enhanced model was tested against the client's current decisioning approach.

Cut-off optimisation

Approval strategies and score thresholds were refined to match risk appetite.

Deployment

The solution was integrated into production decisioning
flows.

PERFORMANCE DETAIL

Side-by-side comparison of bureau-only baseline vs. bureau plus Credolab on the
same population, same observation window.

Metric
Before
After
Default Rate
 11.4%
8.2% (−28%)
Gini
48
56 (+8)
Approval Rate
27%
28% (+4%)
Higher profitability per approved loan
Stronger thin-file assessment
Richer data variables and feature engineering
Improved predictive power and model lift
Real-time decisioning support
Fraud signal detection via device intelligence

Case Study Sheet

Get the Full Case Study

Go beyond the preview and explore:
Full implementation details | Results & business impact | Expert consultation included

More Case Studies

¡Gracias! ¡Tu envío ha sido recibido!
¡Ups! Algo salió mal al enviar el formulario.