MSME credit underwriting in India has been the credit problem the formal banking sector could not crack for two decades. Information asymmetry was severe, credit histories were thin, and traditional underwriting was too slow and too expensive to serve the segment economically. The combination of account aggregator data, GST data, and alternative scoring models has changed this. The default numbers from Q1 2026 are the first quarterly evidence that the new approach is working at scale.

What the data is showing

Default rates on MSME loans underwritten through the new data-rich pipeline are running 30% to 50% below the equivalent vintage from traditional underwriting. The absolute numbers are still higher than retail credit, as expected, but the trend is clearly downward.

Time to decision has compressed from weeks to hours for the standard product. Cost to underwrite has fallen by a factor of 5 to 10 depending on the lender. The unit economics are working.

What changed in the data

Account aggregator framework adoption has reached the threshold where most MSMEs that are willing to share data can be onboarded quickly. GST data is now machine-readable in standard formats. Payment data from the major aggregators is accessible through standard APIs. The four or five data sources that matter for MSME underwriting are no longer manual to assemble.

The models that consume this data are nothing exotic — gradient boosted trees, mostly, with some neural network experimentation at the larger NBFCs. The technology is well-understood. The data was the constraint.

What is still hard

Underwriting MSMEs that operate primarily in cash, or that are very early in their formalisation journey, is still difficult. The data-rich pipeline only works for businesses that have left a digital footprint. Estimates put the addressable market with the new pipeline at around 60% of the total MSME universe.

The other 40% is being served by traditional methods, by NBFC field-collection models, and by informal lenders. The new approach has not yet reached them.

What the lenders are doing differently

The lenders winning in this market have built data engineering and ML capability that traditional banks have lagged on. The pure-play digital NBFCs are leading. Several private sector banks have built equivalent capability through partnerships or acquisitions. The PSU banks have been slower.

The competitive moat is the data pipeline and the model maintenance, not the underwriting decision itself. Lenders that have not invested in the pipeline cannot replicate the economics.

The customer experience

For MSMEs that fit the new model, the loan application experience is genuinely better than it was. A pre-approved offer based on shared data can land in minutes. Disbursement happens in hours, not weeks. The receivables financing products built on the same pipeline have similar improvements.

The transformation in customer experience is what will drive the next phase of adoption. The MSMEs that have access to this experience are not going back to the old model.