Lenders with a book on another system
Sync by CSV or integration and get scoring, queues and strategies without moving the loan book.
Recover · Bankify module
Loan collection software that tells you which account to call first, and why
Collections teams rarely lack data — they lack an order to work in. Bankify scores every delinquent account daily on three signals, ranks the queue by what is actually recoverable, and runs the routine chasing automatically so agents spend their time on the accounts where a conversation changes the outcome. Every score is inspectable, because a collections decision you cannot explain is one you cannot defend.
Rules-based risk scoring you can explain to a credit committee
Every account gets a score from 0 to 100 built from three weighted signals, and the feature values behind each score are stored alongside it. That means an agent, a manager or an auditor can ask why an account scored what it did and get an answer, which is not true of a model that only emits a number.
- Days past due, weighted most heavily, measured against a 120-day horizon
- Balance utilisation — outstanding against original principal
- Payment recency — how long since anything was paid, with never-paid treated as maximally stale
- Bands of low, medium, high and severe, at fixed and published thresholds
- The feature values behind every score are persisted, so any score can be reconstructed later
A queue ranked by what is actually recoverable
Risk alone is the wrong sort order — a severe account with a tiny balance is not the best use of the next hour. The queue is ranked on a separate priority score that combines risk with the balance at stake and how recoverable accounts in that ageing band historically are, so the top of the list is where the recoverable money is.
- Priority ranking that weighs risk, outstanding balance and expected recovery together
- Ageing buckets of current, 1–30, 31–60, 61–90 and 90+ days
- Account statuses of performing, delinquent, default, closed and written off
- Days past due, bucket and status recomputed on every sync rather than trusted from the source system
- Assignment to an agent, with region and segment tags for routing
Automated, DPD-triggered recovery strategies
A strategy is an entry rule plus an ordered set of steps, each with its own days-past-due threshold and channel. Accounts that match the entry rule progress through the steps automatically as they age, so routine chasing happens on time without anyone remembering to do it.
- Entry criteria on bucket, risk band, status, minimum balance and a days-past-due range
- Ordered steps, each firing at its own days-past-due threshold
- Outbound WhatsApp, SMS and email, plus agent tasks for human escalation
- Message templates with the customer name, outstanding balance, days past due and next amount due
- Each step fires at most once per account, enforced at the database rather than in application logic
Contact guardrails, applied to every automated send
Automated collections is one bug away from messaging a customer six times before breakfast. Every automated send — from strategies and campaigns alike — passes the same guardrail check first, and the settings are per-institution. Interactions an agent logs by hand are exempt, because a human already made that judgement.
- Opt-out honoured across every channel
- Quiet hours, defaulting to 21:00–07:00
- A frequency cap defaulting to one automated contact per account per day, across all channels combined
- Configurable per institution, and applied identically to strategies and campaigns
Promise-to-pay, reconciled rather than remembered
A promise to pay is only useful if something checks whether it was kept. Bankify reconciles promises against actual payments nightly using a deterministic rule, so a broken promise surfaces on its own instead of waiting for an agent to follow up.
- Promises captured by an agent or by the bot
- Only payments made on or after the promise was captured count toward it
- Resolves to kept, partial, broken, open or cancelled
- Reconciled nightly, and on demand when you need the position now
Bring the loan book, wherever it lives
Most institutions adopting collections still run their loan book somewhere else, so ingestion was built for that case first rather than as an afterthought. Portfolios sync directly from Bankify, arrive by CSV with a per-institution column mapping, or are pulled from a REST endpoint.
- Direct sync from Bankify Loan Management, incremental after the first load
- CSV import for accounts and payments, with a per-institution column mapping
- Generic REST pull for other systems
- Accounts keyed on your own external reference, so re-imports update rather than duplicate
- Nightly sync and scoring refresh; strategies evaluated hourly
Recovery analytics — and the numbers we refuse to invent
Analytics cover recovery trend, promise-kept rate, agent productivity, strategy effectiveness and portfolio ageing. Roll rate and cure rate are deliberately absent: those need a history of bucket transitions, and reporting them from current-state data would be a fabricated number rather than a derived one.
- Recovery trend by day
- Promise-kept rate
- Agent productivity from logged interactions
- Strategy effectiveness — sent, skipped and failed per strategy
- Portfolio ageing trend from daily aggregates
Runs on your loan book, wherever it lives
Collection Intelligence isn’t tied to Bankify’s own lending engine. Sync it against loans already running on your existing LMS — via direct integration or CSV import — and get the same risk scoring, queues and automated strategies from day one.
Who it’s for
Microfinance institutions
Thousands of small balances where per-account manual chasing does not pay for itself.
Collections teams under audit
Explainable scores, contact guardrails and a logged history of every automated message sent.
Where this fits in the platform
Institutions running Bankify loan management software get their portfolio synced automatically, with arrears arriving the same day rather than at month end.
Promise-to-pay reconciliation matches against real repayments posted in core banking software, not against a separately maintained payment list.
The customer contact details these strategies rely on are the ones captured and verified during digital loan onboarding.
Collections at small-ticket volume is one of the main reasons institutions evaluate Bankify as microfinance software.
Frequently asked questions
How does Bankify score a delinquent account?
On three weighted signals: days past due, balance utilisation, and how long it has been since any payment. Days past due carries the most weight. The score runs 0 to 100 into bands of low, medium, high and severe, and the feature values behind each score are stored so any score can be reconstructed and explained.
Is the risk scoring a machine-learning model?
No, and that is deliberate. It is a transparent, rules-based calculation with published weights and thresholds. Every score keeps the feature values it was computed from, so a manager or an auditor can ask why an account ranked where it did and get a real answer.
Can Collection Intelligence run against a loan book that is not in Bankify?
Yes. Portfolios sync directly from Bankify, import from CSV with a per-institution column mapping, or are pulled from a REST endpoint. Accounts are keyed on your own external reference, so re-imports update existing accounts rather than duplicating them.
What stops automated collections messaging a customer too often?
Every automated send passes a guardrail check first: opt-out is honoured, quiet hours default to 21:00–07:00, and a frequency cap defaults to one automated contact per account per day across all channels combined. The settings are configurable per institution, and interactions an agent logs by hand are exempt.
Which channels can recovery strategies use?
Outbound WhatsApp, SMS and email, plus agent tasks that escalate to a person. There is no automated dialer — an agent task creates work for a human rather than placing a call.
Does Bankify report roll rate and cure rate?
No. Those require a history of bucket-to-bucket transitions, and the portfolio store holds current bucket rather than that history. Reporting them anyway would mean publishing a fabricated figure, so the analytics deliberately omit them.
The rest of the platform
Want to see this against your own loan book?
Talk to a product specialist about which modules fit your institution today.