Transformation · Micro Finance

Two markets that look alike, and repay very differently.

A microfinance lender chose centre locations on demographics alone. Adding credit bureau history showed that markets which look identical on paper can differ sharply on write-offs.

Microfinance lender
Joint liability group lending
Centre formation
Demographic twins separated by credit history

What the bureau overlay found

Two villages in the same region, compared on demographics and then on credit history.

8.1%
Observed
Historical write-off rate in Village A. Pristine credit health.
35.6%
Observed
Historical write-off rate in Village B, which looks the same on every demographic measure.
Zero
Observed
Active customers the lender held in either village.
74%
Observed
Macro write-off rate in a third market where the lender was already heavily present.

The case

01 · Filters running out

The existing pool kept shrinking

Sourcing filters applied to the known customer base returned fewer names each cycle, with no visibility into open market.

02 · Targets rising

Growth expectations outpaced the method

Annual targets climbed faster than a centre formation process built on local familiarity could reliably serve.

03 · Risk averaged away

Delinquency was read at portfolio level

Write-off behaviour was understood as a portfolio number, which hides the fact that it concentrates geographically.

Demographics said go. The credit history said one of them was a trap.

Both villages are densely populated, commercially active, and hold no customers of the lender. On a population heatmap they are the same recommendation. The bureau overlay separates them completely.

Village A · the unmined market
Residents
~34,900
Historical write-off
8.1%
Dense, commercially active, zero competitor penetration, clean credit history. A genuine blue ocean, and the lender had never entered it.
Village B · the deceptive trap
Residents
~28,100
Historical write-off
35.6%
Same profile on every demographic measure, and a write-off rate four times higher. On a heatmap this village reads as a primary expansion target.
What it means
Group lending survives on small tickets and high recovery. A market with write-offs at this level breaches the tolerance of even a resilient product, so capital deployed there is very likely to be lost. Demographic selection alone cannot see the difference, because the difference is not demographic.

What was scored

Six dimensions per micro-market, each with a defined threshold rather than a judgement call.

Demographic density

Residential density measured against the state median for population and buildings.

A centre needs enough households within walking distance to form a five-member group at all.

Debt burden

Principal outstanding per capita against the state upper quartile.

Distinguishes an under-served market from one where households are already over-leveraged.

Macro credit risk

Historical write-offs and active defaults at the surrounding postcode level.

Sets the absolute ceiling. Past this, no group loan is viable regardless of demand.

Own portfolio behaviour

The lender's own arrears rate inside the micro-market.

Separates a market that is generally difficult from one where this lender specifically is struggling.

Footprint

Active customers of the lender inside the cell.

Identifies genuinely virgin territory, which is where first-mover economics apply.

Competitive intensity

Banks, other lenders and other microfinance institutions inside the 400 metre block.

Group lending saturates quickly, so competitor count at close range predicts over-leverage.
Six scored dimensions per micro-market

Six classes, and one of them says stop

Opportunity, debt burden, risk, footprint and competition combine into a single instruction per micro-market.

Class
What it means
Action
Aggressive Blue Ocean
Dense population, low debt burden, clean credit, no formal competitors.
Form new groups immediately
Fortify and Harvest
Existing groups performing, collections strong, credit clean.
Renew and cross-sell
Cautious Skimming
High density with elevated historical defaults.
Verify members individually, never on group pressure
Early Warning
Active groups present in a market whose credit behaviour is deteriorating.
Restructure before it spreads
Slow-Burn Digital
Viable but too dispersed for a field officer to serve economically.
Serve digitally, do not deploy
Total Freeze
Postcode write-off history has breached maximum tolerance, whatever the demand looks like.
Halt all new group formation
Six classes across a district

What each team received

Deliverables arriving on different cadences.

Leadership

Location questions in plain language

A chat interface over the branch data, so leadership can ask a location-grounded question without needing mapping skills or an analyst in the loop.

Branch managers

A ranked micro-market shortlist

Hexagons ordered per branch with a go or reconsider flag, so centre formation starts from evidence rather than from where the last centre happened to be.

Risk and field

Stop-sourcing zones

A hexagon-level risk overlay combining the lender's own arrears with bureau delinquency, naming the markets where new group formation halts.

How the workflow changed

Before
New centres formed near existing ones
Sourcing filters applied to a shrinking known pool
Write-off risk read as a portfolio average
Competitor saturation assessed by local impression
No formal way to rule a market out
→
After
Centres formed against a ranked micro-market list
Open market surfaced alongside the existing base
Write-off risk carried at micro-market resolution
Competitor count measured inside a 400 metre block
A stop-sourcing classification with a defined threshold

How the engagement was gated

Three stages, with a decision point that could end it.

Month 1
Model built across the pilot branches, with territory maps, risk overlays and beat plans delivered.
Setup
Gate
Branch managers confirm the classification matches what they know of their own markets.
Months 2 to 3
Officers deployed against the beat plans, monitoring alerts live, risk zones enforced.
Monitoring
Gate
Field teams report whether territory quality and planning time actually improved.
End of month 3
Joint leadership review of the pilot branches.
Go or stop
Gate
Early results and field adoption decide whether the engagement scales or ends.
The filtered customers we are getting now keep reducing. It just keeps reducing.
Client Leadership
On the existing sourcing method

A filter applied to a known base can only shrink. The way out is to see the open market, and the reason that had not been done is that open market and unsafe market look the same until credit history is placed on the map.

Test it on a market you already rejected.

Name a market your teams walked away from and one they are keen on. We will classify both and show you the credit history underneath the demographics.

Set up a workshopMore Transformation Examples
EPIC Intelligence · Client identity withheld