Turn delinquency into predictable recovery.
Crux AI Collection Scorecard predicts which overdue accounts will cure, routes each one to the right strategy, and puts your agents on the cases where a call changes the outcome. Built for banks, finance companies and fintechs in Saudi Arabia.
What is an AI collection scorecard?
Rather than treating every overdue account the same, a collection scorecard scores each one individually, so effort goes where it changes the result.
That lets a lender concentrate agent time on cases where a conversation matters, automate outreach for mid-risk accounts, and avoid unnecessary contact with customers who are likely to pay on their own.
Every account gets the right treatment, automatically
Each account is segmented by cure probability, balance and behavioural history, then routed to one of four strategies in real time. Select a lane to see how it works.
// overdue accounts, scored and sorted
High-probability accounts are watched, not chased. That protects the customer relationship and removes contact cost that would not have changed the outcome.
Strategy thresholds and treatments are configured with your collections and risk teams. Nothing is hard-coded.
Seven capabilities, one collections brain
Cure probability scoring
Dynamic 30, 60 and 90-day cure probabilities for every account, refreshed daily as new repayment signals arrive. This is the predictive core that drives every strategy decision.
- 30, 60 and 90-day cure windows
- Daily refresh with new behaviour
- Segment-specific models
- Confidence reported per account
Agent workload optimisation
Agent queues ranked by expected value so people spend their day on accounts where a conversation changes the outcome, not on cases that will resolve themselves.
- Expected-value ranked queues
- Skill-to-account matching
- Best-time-to-call prediction
- Call outcome learning loop
Restructuring offer engine
Recommends the arrangement a customer is most likely to accept, balancing recovered value against the customer's affordability signals.
- Affordability-based offer sizing
- Repayment plan optimisation
- Settlement propensity scoring
- Offer acceptance prediction
NPL portfolio analytics
Dashboards for flow rates, cure rates by vintage, provision coverage and recovery cost, so portfolio owners see what is working and where the leakage is.
- Roll-rate analysis
- Vintage cure tracking
- Provision adequacy monitoring
- Recovery ROI per strategy
Omnichannel contact strategy
Chooses the channel, timing and message for each account (SMS, email, app or call) based on how that customer is predicted to respond.
- Channel preference prediction
- Optimal timing per account
- Personalised message set-up
- Response tracking and learning
Compliance & fair treatment
Vulnerability flags, over-indebtedness detection and contact-frequency limits, with a complete interaction log that helps a lender evidence fair treatment to SAMA.
- Vulnerability indicators
- Fair-treatment audit trail
- Contact frequency controls
- Full interaction log
Portfolio stress simulation
Simulate macroeconomic stress, repayment shocks and behavioural shifts to forecast recovery curves and provisioning impact before they hit the book.
- Macroeconomic scenarios
- Repayment shock simulation
- Recovery curve forecasting
- Forward-looking provisioning
Rule-based collections vs AI collections
Improvement in recovery and cost-to-collect varies by portfolio. We agree measures and a baseline with you before rollout and report against them.
Part of the full credit lifecycle
The Collection Scorecard connects with the rest of the Crux platform, from origination through portfolio monitoring to recovery and execution.
Build credit scorecards that help you decide who to approve.
Explore → MonitoringBehavioural ScorecardTrack existing accounts and catch rising risk before it reaches collections.
Explore → RecoveryCollection ScorecardPredict who will cure and choose how to treat each account.
You are here ExecutionAI Decision EngineRun every scorecard output through a no-code rules engine and automate workflows.
Explore →Answered plainly
What is an AI collection scorecard?
A predictive model that estimates the probability that a delinquent account will cure, using repayment behaviour, customer characteristics and macroeconomic signals. Unlike static rule-based collections, it updates as new data arrives and uses the score to choose the most cost-effective strategy for each account.
How does AI improve debt recovery?
AI predicts which accounts will self-cure, which respond to digital outreach, which need an agent, and which should move towards legal action. Effort is then concentrated where it changes the outcome, instead of spreading the same treatment across every overdue account. Results depend on the portfolio, and the improvement is measured against your current approach during rollout.
What is cure probability modelling?
It predicts how likely a delinquent account is to return to performing status within 30, 60 or 90 days without restructuring or legal action. The model analyses repayment patterns, behavioural signals, balance characteristics and economic data to give each account a dynamic score.
Does the platform support SAMA fair-treatment expectations?
It includes vulnerability flags, over-indebtedness checks, contact-frequency controls and a complete interaction audit log, which help a lender evidence fair treatment of customers. Your compliance team remains responsible for policy and for interpreting SAMA requirements.
Can it work with our existing collections system?
Yes. The scorecard reads account and repayment data from core banking and your collections system through APIs or batch files, and returns scores, strategies and queue rankings to the tools your agents already use. It can also feed the Crux Decision Engine for no-code workflow automation.
Turn collections into a competitive advantage
In a focused 30-minute strategy session we review your delinquency data, show how accounts would be segmented, and outline what a first release and its success measures could look like.