Customer story

A leading bank Up to 15% fewer losses on its riskiest business customers, at the same approval rate.

A leading bank in Brazil tested Monnai on its business customers. Monnai added business-network, ownership and digital-presence signals to the bank’s applications and rebuilt the model on the bank’s own lending outcomes.

  • Up to 15%fewer losses in the riskiest band, at the same approval rate
  • 12–20%more approvals in the riskiest band, at the same loss rate
  • 31%of businesses in the riskiest grade defaulted, against under 1% in the safest

Measured on the bank’s own lending outcomes, including months the model had never seen.

Where Monnai sharpens the bank’s score the most. Gain in how well the score separates businesses that repay from those that default, on months the model had never seen.
  • Thin-file, higher-risk businesses+21%
  • Health services+17%
  • Auto trade and repair+12%
  • Sole proprietors+6%
  • All businesses+5%
Default rate by risk grade. The bank’s score and Monnai combined, in eight grades of about 10% to 15% of businesses each.
  1. AUnder 1%
  2. B1%
  3. C3%
  4. D4%
  5. E7%
  6. F11%
  7. G18%
  8. H31%
New businesses, decided rather than turned away. Monnai’s decision for businesses less than a year old, about half of the applicants in the test.
  • Approve41% automatically
  • Review or collateral47%
  • Decline, with reasons12%
More warning signs, a higher risk tier. Share of businesses rated high or critical risk, by the number of adverse markers in public records: credit protests, tax debt, bounced checks.
  1. None4%
  2. One39%
  3. Two78%
  4. Three or more91%

Charts from the customer’s proof-of-concept reports, anonymized and rounded.

The challenge

The bank’s own score ranked the safe majority of its business customers well, but struggled with the riskiest. Sole proprietors, a fifth of the book, default almost twice as often as average, often with little credit history to go on.

What Monnai did

  • Added business-network, ownership and digital-presence signals to every business application.
  • Rebuilt the model on the bank’s real lending outcomes, and tested it on months it had never seen.
  • Blended it with the bank’s own score, so every decision keeps the bank’s view and adds Monnai’s.

What this means for credit and risk teams

  • Fewer losses where they hurt: up to 15% fewer losses in the riskiest band, at the same approval rate.
  • More approvals, same losses: 12–20% more approvals in the riskiest band, at the same loss rate.
  • Grades a credit committee can use: from under 1% defaults in the safest grade to 31% in the riskiest.
  • Clear drivers: businesses with no digital presence default about twice as often; the number of business partners and the sector matter too.

More customer stories

All customer stories

Let us prove it. Test us on your data.

Send past applications or transactions. We score them and send back a full report: every score with its reasons, and the uplift for your portfolio.

5 business days to your full report. We commit to it.

Start your testRequest a call back