Independent decision analytics for lenders

Find the decisions quietly costing your lending book.

I help NBFC leadership teams replay historical pricing, renewal, approval, collections and channel decisions—then test transparent challenger rules that target incremental contribution profit without loosening risk guardrails.

Discuss one portfolio decision → See the advisory model

A bounded diagnostic first. No production change, black-box replacement or transformation programme required.

01 Historical decisions replayed

02 Transparent rules, not mystery scores

03 Profit measured after risk and cost

04 Audit-ready handoff

Where value tends to hide

Four decisions worth replaying.

The work starts where the institution already makes repeatable decisions—and where small policy changes can compound across a portfolio.

Pricing and approval

Which approved loans were under- or over-priced for realised risk and cost?

Replay booked cohorts to locate pricing bands, exception rules and approval cut-offs that suppress contribution profit.

Output: cohort-level pricing and exception rules

Renewal and limits

Who should receive which limit, price and timing?

Use customer behaviour, tenure and repayment outcomes to challenge blunt renewal rules without weakening portfolio safeguards.

Output: renewal decision tree and challenger policy

Collections routing

Which account should receive which treatment, when and through which channel?

Compare treatment sequences against net recoveries, contact costs and operational capacity—not activity counts alone.

Output: treatment priorities and capacity plan

Dealer and channel allocation

Which partners create durable risk-adjusted value?

Separate volume from contribution by attributing downstream losses, incentives and collections cost to the originating channel.

Output: channel scorecard and sourcing rules

A challenger before a change

Evidence first. Policy second.

A controlled diagnostic turns historical decisions into an auditable comparison between the current policy and one carefully bounded alternative.

Frame one decision

Choose one portfolio, product or geography and agree the business objective, constraints and decision owner.

Reconstruct reality

Join decisions, prices, outcomes, losses, costs and exceptions into a cohort-level view of what actually happened.

Replay the policy

Measure the current rule, surface leakage and test transparent challengers against the same historical population.

Hand back rules

Deliver implementable bands, trees or priorities with monitoring metrics, governance notes and a champion-versus-challenger plan.

The bounded pilot

Small enough to approve. Concrete enough to judge.

Start with one recurring decision where data exists, outcomes can be measured and the internal owner wants a sharper answer.

1Decision or portfolio scope

4–6Weeks for the diagnostic

0Production changes required

Designed for scrutiny

Not another black-box model pitch.

The work is built to survive questions from risk, finance, product, operations and governance—not just impress a data team.

Commercially grounded

Contribution profit, not approval growth in isolation

Value is calculated after funding, losses, incentives and operating costs wherever the available data permits.

Operationally usable

Rules a business team can understand and implement

The handoff favours bands, decision trees and prioritisation logic over opaque complexity.

Governance aware

Guardrails and exceptions made visible

Every challenger is evaluated alongside agreed loss, approval, fairness and policy constraints.

Independent

No incentive to sell a platform replacement

The objective is to clarify the decision and leave behind a useful policy—not create dependency.

A clear boundary

“The first conversation is not about replacing your underwriting stack. It is about identifying one decision where an independent historical replay could reveal material value.”
Dwaipayan Chakraborty — Independent advisor