Collections Optimization
“Which delinquent accounts will self-cure vs need intervention?”
Book a demo and get a free trial of the full platform: data science agent, fine-tune capabilities, and forward-deployed engineer support.
By submitting, you accept the Terms and Privacy Policy.
Loved by data scientists, ML engineers & CXOs at
A real-world example
Which delinquent accounts will self-cure vs need intervention?
Banks spend $2-4B annually on collections operations, yet 40-60% of early-stage delinquent accounts self-cure without any intervention (McKinsey). Treating every 30-day-late account with the same urgency wastes collector time on accounts that would have paid anyway while under-prioritizing accounts heading toward charge-off. A bank with 200K delinquent accounts per month needs to know which 80K genuinely need a call, which 40K need a workout plan, and which 80K will resolve on their own.
How KumoRFM solves this
Relational intelligence built for banking and financial data
Kumo connects delinquent accounts to their full payment history, transaction patterns, employment signals, other credit lines, and prior collections outcomes. The model learns that Account L-8002 missed a payment but has consistent direct deposits, no balance growth on other lines, and a history of catching up within 15 days. Meanwhile, Account L-8045 shows declining income signals, rising utilization across all cards, and a pattern of minimum-only payments. Kumo routes collectors to the 40% of accounts that truly need intervention.
From data to predictions
See the full pipeline in action
Connect your tables, write a PQL query, and get predictions with built-in explainability — all in minutes, not months.
Your data
The relational tables Kumo learns from
DELINQUENT_ACCOUNTS
| account_id | borrower_id | product | days_past_due | balance_owed |
|---|---|---|---|---|
| L-8002 | B-2087 | Personal Loan | 32 | $14,200 |
| L-8045 | B-2120 | Auto Loan | 45 | $22,800 |
| L-8067 | B-2155 | Credit Card | 31 | $6,400 |
PAYMENT_HISTORY
| account_id | month | amount_due | amount_paid | days_late |
|---|---|---|---|---|
| L-8002 | 2025-07 | $310 | $310 | 0 |
| L-8002 | 2025-08 | $310 | $310 | 3 |
| L-8045 | 2025-07 | $485 | $485 | 0 |
BORROWER_SIGNALS
| borrower_id | direct_deposit_trend | total_utilization | other_delinquencies |
|---|---|---|---|
| B-2087 | Stable | 38% | 0 |
| B-2120 | Declining -15% | 87% | 2 |
| B-2155 | Stable | 52% | 0 |
PRIOR_COLLECTIONS
| borrower_id | prior_delinquency | outcome | days_to_resolve |
|---|---|---|---|
| B-2087 | 2024-03 | Self-cured | 12 |
| B-2120 | 2024-11 | Workout plan | 90 |
| B-2155 | None | N/A | N/A |
Write your PQL query
Describe what to predict in 2–3 lines — Kumo handles the rest
PREDICT BOOL(DELINQUENT_ACCOUNTS.STATUS = 'self_cured', 0, 30, days) FOR EACH DELINQUENT_ACCOUNTS.ACCOUNT_ID WHERE DELINQUENT_ACCOUNTS.DAYS_PAST_DUE > 30
Prediction output
Every entity gets a score, updated continuously
| ACCOUNT_ID | BORROWER | SELF_CURE_PROB | RECOMMENDED_ACTION | PRIORITY |
|---|---|---|---|---|
| L-8002 | B-2087 | 0.82 | Monitor Only | Low |
| L-8067 | B-2155 | 0.61 | Soft Reminder | Medium |
| L-8045 | B-2120 | 0.11 | Collector Outreach | Critical |
Understand why
Every prediction includes feature attributions — no black boxes
Account L-8045 (Auto Loan, B-2120)
Predicted: 11% self-cure probability (needs intervention)
Top contributing features
Income signal declining
-15% deposits
28% attribution
Cross-account utilization
87% total
25% attribution
Multiple concurrent delinquencies
2 other
21% attribution
Prior collections required workout
90 days
15% attribution
Days past due trajectory
Worsening
11% attribution
Feature attributions are computed automatically for every prediction. No separate tooling required. Learn more about Kumo explainability
PQL Documentation
Learn the Predictive Query Language — SQL-like syntax for defining any prediction task in 2–3 lines.
Python SDK
Integrate Kumo predictions into your pipelines. Train, evaluate, and deploy models programmatically.
Explainability Docs
Understand feature attributions, model evaluation metrics, and how to build trust with stakeholders.
Bottom line: Focus collector effort on the 40% of delinquent accounts that genuinely need intervention, reducing collections costs by $800M-$1.2B industry-wide while cutting charge-off rates by 15-20%.
Related use cases
Explore more financial services use cases
Topics covered
One Platform. One Model. Predict Instantly.
KumoRFM
Relational Foundation Model
Turn structured relational data into predictions in seconds. KumoRFM delivers zero-shot predictions that rival months of traditional data science. No training, feature engineering, or infrastructure required. Just connect your data and start predicting.
For critical use cases, fine-tune KumoRFM on your data using the Kumo platform and Data Science Agent for 30%+ higher accuracy than traditional models.
Book a demo and get a free trial of the full platform: data science agent, fine-tune capabilities, and forward-deployed engineer support.
