Breaking Down Loan Business Using First Principle Thinking
When you strip the loan business down to its core, three fundamentals stay:
The Borrower’s Ability & Willingness to Repay
Will the money come back or not?
Cashflow capacity
Business volatility
Seasonality & stability
Willingness to pay
Fraud risk
The Terms You Offer
Under what condition should the borrower repay?
Repayment schedule
Tenor
Installment size
Collateral
Triggers & consequences
The Price of the Credit
What does the borrower pay to use the money?
Interest
Fees
Risk premium
Margins
Cost of capital pass-through
Why MSME Lending in Indonesia is Hard?
To evaluate creditworthiness, design the terms, and set the price, lenders need clean, consistent data. MSME lending in Indonesia is tricky specifically because so many data are missing. Here’s the reality on the ground:
Most of the transactions are still cash-based.
Non existent book keeping means loss of data. Information about cash flow, revenue, cost, profit, and margins is missing. Details on working capital needs and payment schedule are also gone.
Most of them mix personal and business money.
Loans are obtained through informal channels. They can’t be traced reliably. Thus, we can’t fully understand how much total debt the business has.
Two Paths to Solving the Data Gap
They way I’m seeing this, there are two ways to crack the problem of missing data:
Path 1 – The Long Road: Digitalize Transactions
Shifting merchants away from cash sounds great on paper, but the friction is real:
Cash is tangible, instant, universally accepted.
Non-cash payment method still depend on an ecosystem that isn’t fully ready.
Efforts like Mitra Bukalapak, BukuWarung, and others helped—but haven’t fundamentally changed behavior at scale.
This route will happen eventually, but it will take years, not quarters.
Path 2 – The Practical Shortcut: Infer Data from the Ecosystem
Here’s the part most lenders miss:
You don’t need the merchant to provide data. You need data about the merchant.
For example, a large distributor supplying thousands of small retailers already has:
Order frequency
Basket size trends
Seasonality patterns
SKU velocity
Growth or contraction signals
This dataset alone can estimate what MSMEs can’t (or won’t) record: their operational rhythm and sales trajectory.
Stop waiting for primary data. Start mining secondary data.
What It Takes to Lend to MSMEs Today
To operate in today’s messy data environment, lenders need to do two things:
Collect Indirect Data Points That Can Be Mapped Back to Each Merchant
Distributors, wholesalers, POS networks, logistics providers, payment gateways—these are all reservoirs of behavioral patterns.
Make Sense of Noisy Data Using Machine Learning
When the data is partial, inconsistent, or indirect, you need ML models to:
Detect patterns
Infer business health
Estimate stability
Predict repayment capacity
It’s the only scalable method to navigate messy, incomplete datasets.
The Core Insight
Penetrating Indonesia’s MSME lending market requires: leveraging existing external data + using machine learning to cut through noise.
Clean ledgers and structured financials won’t magically appear anytime soon. But the signals around the business are already there.
What’s Next
In the next post, I’ll explore how to systematically identify these “hidden data points” using Systematic Inventive Thinking (SIT).
One example you’ve already seen is Task Unification: Repurposing distributor data to provide operational clarity on thousands of MSMEs.
We’ll also look at the other four SIT tools:
Subtraction
Multiplication
Division
Attribute Dependency
Each one opens up new ways to solve the MSME data scarcity problem—without waiting for the entire ecosystem to modernize.
