Insights
Worked cases from Luminnovate's research. Each one answers a question a credit, risk or finance team asks, and is labelled by the data it comes from.
- Revenue overstated by inflowsHow much of what comes into an SME account is really revenue?Illustrative worked example
- The builder whose loan masked a fallCan a steady looking account hide a falling business?Synthetic business
- The contractor and the cafeIs an irregular cash flow a risk, or just how the business works?Synthetic business
- Overdraft fees before missed repaymentsDoes the bank account warn of default before the first missed repayment?Public bank data
- Risk or mix?Did borrowers get riskier, or did we lend to different borrowers?Public lending data
- The collections fix that looked like a modelWhen late payments collapse, what actually changed?Anonymised invoice data
- A fall that was noiseWhen revenue falls, when should you not act?Public retail data
- The simpler model that matchedDoes a decision need a complex model?Public lending data
- The number that passed the checkHow do you stop AI inventing a figure?Luminnovate's own testing
- Peers cut false alertsDoes judging against peers work outside lending?Public energy data
- Seeing missed repayments comingDoes early warning beat the rule a lender already runs?Public bank data
- Ranking accounts, not paymentsHow should a fixed review team choose which accounts to hold?Synthetic bank transfers
- Which hardship plans holdIs the reason for a hardship plan a signal of what comes next?Public lending data
- Proving an alert is worth sendingHow do you know acting on an alert changed anything?Simulation
- A book under every scenarioCan a projection of the book come with ranges you can rely on?Simulation
Data sources are available on request; methods are proprietary.
Book a conversation