A rural lender ran one lending model across every branch. Each branch was matched to the model its surrounding market can actually sustain, and the results were checked against field data before any rollout.
Phased deliberately so that technical validation happens before large capital commitment.
Group lending and enterprise lending were pushed through the same branches without distinguishing which markets suited which.
A prosperous commercial cluster looks like a prime target on any demand measure, and it is often unsuitable for group lending precisely because it is prosperous.
Deploying across several hundred branches without a validation stage puts capital behind a model nobody has checked on the ground.
The same signals that make a micro-cluster attractive for enterprise lending can rule it out for group lending. High affluence proxies, dense registered businesses and strong road connectivity point one way. Group lending needs the opposite profile.
Residential density sufficient to form groups, low affluence proxies, thin formal competition. The branch leads with group lending.
High-street ecosystem, dense registered businesses, strong connectivity. Group lending is unsuitable here and enterprise lending is the play.
Both profiles present in different parts of the same service area, which is a territory design question rather than a product choice.
A worked example makes the method concrete. One micro-cluster in the pilot carried the following profile.
Over 70 registered entities, weighted toward apparel manufacturing, pharmacy and vehicle showrooms.
Rated 4.8 out of 5, driven by concentrated apparel manufacturing.
Over 450 structures carrying commercial height profiles.
Three non-bank lenders and two cash points within one kilometre.
This cluster scored as a primary enterprise lending target and was ruled out for group lending, on the same evidence.
Each answers a different question, and each names the action.
The rationale matters as much as the rank. A hotspot listed as a dense textile cluster and one listed as an under-banked agricultural hub call for different conversations when the officer arrives.
Deliverables arrive in sequence, matched to what each phase is trying to establish.
In the final phase, network-wide performance tracking with alerts raised to zonal level rather than waiting for a review cycle.
The classification that sets which product to lead with, and the ranked micro-clusters inside the service area to work first.
The second phase measures adoption speed directly, meaning how quickly teams pivot to the identified hotspots.
A joint field audit designed to test the contrast, rather than to confirm the winners.
That objection is the reason the audit is joint and the gate is explicit. Validation is not left to the client to fund and run alone, and the rollout does not proceed on assertion. If the maps do not match the ground at the agreed rate, the engagement stops there.
Give us the branches in one region. We will return an archetype for each and you can audit the ones you disagree with first.
Set up a workshopMore Transformation Examples