From Data Gathering to Due Diligence

The Missing Layer in MSME Lending

Walk into most credit departments and observe how a loan application progresses.

A surprising amount of time is spent not on analysing the borrower’s business, but on collecting the information required to analyse it.

Statements are requested. GST returns are downloaded. Bank statements are parsed. Credit bureau reports are pulled.

An entire ecosystem of digital tools now exists to make this process faster.

Yet an uncomfortable question remains:

Has the quality of due diligence improved in proportion to the amount of data being gathered?

The Structure of a Loan Decision

In principle, every lending decision follows a clear sequence.

1. Information gathering

2. Due diligence

3. Application of business rules aligned with risk appetite

4. Creation of the credit offer

5. Acceptance and disbursement.

However, modern lending systems have introduced a sophisticated operational layer between these stages.

In many institutions today, the workflow looks more like this:

Information Gathering → Data Aggregation → CCV Algorithms → RAG Coding → Product Classification → STP Journey or Manual Underwriting → Offer Creation → Acceptance → Disbursement.

Due Diligence Cycle

This architecture enables scale, speed and portfolio discipline.

But it also shapes how risk is interpreted.

The Role of CCV Algorithms (Due Diligence)

Most digital lending frameworks rely on Credit Control Validation (CCV) algorithms.

These algorithms evaluate whether a borrower satisfies predefined policy parameters such as:

• financial ratios

• turnover levels

• bureau behaviour

• transaction routing patterns

• compliance indicators.

In essence, CCV algorithms answer a simple question:

Does this borrower satisfy the eligibility rules for a particular lending product?

They are typically deterministic rule engines designed to ensure policy compliance and operational consistency.

However, they do not interpret the deeper economic context of the borrower’s business.

RAG Coding and Risk Visualization

Many institutions supplement CCV frameworks with RAG coding — Red, Amber and Green indicators.

• Green indicates parameters comfortably within risk thresholds.

• Amber signals borderline conditions requiring review.

• Red identifies parameters outside acceptable limits.

RAG coding allows credit teams to quickly prioritize attention and triage cases.

But it is important to remember that RAG coding does not replace due diligence.

It simply indicates where deeper analysis should begin.

Product Architecture: Templated vs Non-Templated Lending

Modern lending systems also distinguish between templated (schematic) and non-templated products.

Templated or Schematic Products

These products are designed for scale and standardization.

Examples include:

• GST-linked lending

• supply chain financing

• program-based MSME lending

• small-ticket credit products.

Such products rely heavily on CCV algorithms and often enable Straight Through Processing (STP) journeys.

Non-Templated Products

These include:

• bespoke working capital facilities

• structured credit exposures

• project financing

• complex MSME lending.

These cases require manual underwriting and deeper credit judgement.

The FIT Rank Question

Some lending systems also attempt to generate what is sometimes called a FIT Rank — an internal indicator of how well a borrower aligns with the structural and risk characteristics of a particular lending product.

In theory, this helps determine whether a borrower should move through straight-through processing, require manual underwriting, or be considered unsuitable for a given product framework.

However, many practitioners feel that such constructs are not always practical in the Indian MSME context.

Why FIT Rank Is Difficult in MSME Lending

The challenge lies in the nature of the MSME ecosystem itself.

A large proportion of Indian MSMEs operate as:

• proprietorships

• partnership firms

• closely held family businesses.

These businesses often exhibit characteristics that make rigid fitment scoring difficult:

• fluctuating financial patterns

• informal supplier networks

• heavy dependence on promoter behaviour

• evolving business models.

Two businesses with very similar financial metrics may exhibit very different risk characteristics once their operating realities are understood.

In such environments, fitment scores derived purely from structured datasets may not fully capture the qualitative aspects of credit risk.

Where FIT Rank Works Better

Fitment frameworks tend to work better in environments where:

• financial reporting is standardized

• governance structures are formalized

• operating models are stable and transparent.

These conditions are more common in large corporate lending or mature SME ecosystems.

The Real Determinant

For MSME lending, experienced bankers often rely on a combination of:

• financial analysis

• behavioural indicators

• cluster knowledge

• promoter credibility.

In other words, the real determinant of credit quality often lies not in how well a borrower fits a model, but in how well the lender understands the borrower’s business.

In many MSME lending situations, risk does not arise because the borrower fails to fit the model. It arises because the model fails to capture the borrower’s reality.

The Rise of STP Journeys

For templated products, many institutions now deploy Straight Through Processing (STP) journeys.

In STP systems:

• data is automatically fetched from multiple sources

• CCV algorithms assess eligibility

• RAG indicators highlight potential issues

• approvals are generated within defined parameters.

This dramatically improves processing speed.

However, STP systems rely heavily on a key assumption:

that available data signals accurately represent the borrower’s underlying economic reality.

The MSME Challenge

This assumption becomes particularly complex in MSME lending, especially for proprietorship and partnership firms.

Unlike large corporates, these entities often lack:

• robust audited financial statements

• formal governance structures

• detailed disclosure frameworks.

In many cases, the most meaningful risk signals lie outside formal datasets, such as:

• promoter conduct

• supplier and buyer relationships

• industry cluster dynamics

• working capital behaviour.

Yet most digital tools focus primarily on aggregating data rather than interpreting these deeper signals.

Institutional Diligence Ecosystems

Some leading banks have attempted to bridge this gap by building structured diligence ecosystems.

Rather than relying only on publicly available data, these institutions systematically capture insights gathered through lending activity over time.

These repositories may include information relating to:

• promoter reputation within trade networks

• supplier and buyer relationships

• behavioural patterns across prior borrowing cycles

• cluster-level intelligence about industries and markets.

Over time, such systems evolve into institutional diligence banks, where accumulated insights become available to credit teams evaluating new proposals.

In effect, the organization develops a memory of the MSME ecosystem it serves.

The Individual Knowledge Problem

In many banks, however, valuable MSME intelligence does exist — but it often resides with individuals rather than systems.

Certain experienced bankers painstakingly gather information over years from sources such as:

• trade associations

• industry clusters

• informal market networks

• supplier ecosystems

• past borrower behaviour.

They build deep understanding of local business environments.

But this creates another challenge.

The knowledge often remains unstructured and difficult to retrieve.

The key to that information becomes the individual who holds it.

When Knowledge Becomes Person-Dependent

When institutional intelligence resides with individuals rather than systems, several problems arise.

1. Retrievability becomes difficult. Information sharing becomes inconsistent.

2. Institutional learning slows down. Dependency on specific individuals increases.

In practice, teams often know that someone in the organization possesses valuable insight about a borrower or industry cluster.

But locating and retrieving that information may take time.

In time-sensitive credit decisions, that delay can matter.

The Missing Investment

Building structured repositories of MSME intelligence requires deliberate institutional investment.

This involves:

• capturing insights from past credit decisions

• recording behavioural observations

• organizing cluster-level intelligence

• creating searchable knowledge systems.

Yet many banks continue to rely on individual expertise rather than institutional knowledge frameworks.

As a result, insights gathered over decades often remain fragmented across people rather than embedded into decision systems.

From Data Aggregation to Decision Intelligence

The real opportunity in modern credit systems lies in moving beyond:

data aggregation → rule application.

Instead, the next generation of lending platforms must focus on:

data → insight → judgement.

This requires combining:

• automated data aggregation

• CCV frameworks

• structured diligence repositories

• experienced credit judgement.

A Final Thought

Technology has dramatically improved the speed with which banks can gather information.

But the success of lending ultimately depends not on how quickly data is collected, but on how well risk is understood.

The institutions that will succeed in MSME lending will not simply have the fastest STP journeys.

They will be the ones that transform individual intelligence into institutional intelligence.