Every institute owner has a sixth sense for who will pay late. The parent who 'forgot' twice last quarter. The family that needs a phone call, not a reminder. Feezy's AI defaulter predictor turns that sixth sense into a number — and shows you who to call 14 days before the due date.
Signals the model reads
We watch eight signals per active member. Past payment punctuality, average days past due, channel responsiveness (does this parent open WhatsApp?), payment method volatility, recent failed transactions, sibling-account behavior, batch-level cohort timing, and seasonal patterns (school exam weeks, festival cash crunches).
Each member ends up with a risk score between 0 and 100. Above 70 means 'call this person'. Between 40 and 70 means 'send the personalised reminder, not the template one'. Below 40 means 'leave them alone — they'll pay on time'.
What you actually do with the signal
The dashboard shows you a list of high-risk accounts every Monday morning, sorted by expected payment date. Click any name to see the score breakdown and the recommended next action. Most institutes work the top 10% of the list — that's the slice that drives 80% of recovered cash.
One Bengaluru institute reduced its 60-day-overdue book by 67% in the first quarter using nothing but the Monday list. They did not add staff. They did not change their fee structure. They just stopped calling the wrong people.
Privacy and fairness
The model uses only the institute's own first-party data. It does not pull credit scores, demographic data, or cross-institute behavior. Every score is explainable — you can see exactly which signal moved it. And every parent can request the data we hold for them in one click.
