Blog · AutoCollect
Why hardship and disputes should always go to a person in collections.
Automation is good at reminding people who will pay. It is the wrong tool for the borrower whose life has just gone wrong.
Most borrowers in an early bucket are not in trouble. They forgot, they were travelling, the salary landed a day late, the account they pay from was short. For them, a polite reminder and a payment link on WhatsApp is the best possible experience, and an automated one is perfectly fine.
A smaller group is different. A job has been lost, a family member is in hospital, a business has had a terrible season, or the borrower believes the amount being demanded is wrong. Our view is firm: these conversations should always go to a trained person, and automation’s main job is to recognise them early and get out of the way.
Why automation is the wrong tool here
A voice agent or a reminder flow is built around a goal: get a payment or a promise to pay. That goal is right for most accounts. For a borrower in hardship, it can be actively harmful. Pressing for a promise the borrower cannot keep produces a broken promise, another reminder, another call and a borrower who stops answering. Everyone loses: the lender, the borrower and the trust between them.
Hardship also needs judgement that a script cannot carry. Is this temporary or permanent? Is a short deferment enough, or does the account need restructuring under your policy? What can be offered, and what needs approval? Those are decisions for a person with the authority and the training to make them, inside your lender’s policy. For the regulatory side, confirm with your compliance adviser.
Disputes are not objections
A dispute is a borrower saying the amount, the charge or the record is wrong. A script designed to handle objections treats it as resistance to overcome. That is the wrong frame. A dispute is a question of fact, and it needs someone who can look at the account, check the payment history and correct an error if there is one. Until that happens, chasing the disputed amount only creates complaints.
The real job for automation: notice
Automation earns its place here not by handling hardship but by spotting it. Signals worth routing to a person immediately:
- The borrower mentions illness, job loss, death in the family, disaster or business closure.
- The borrower says the amount or a charge is wrong.
- Repeated broken promises with explanations that point to inability, not unwillingness.
- Distress, anger or confusion in the conversation.
- A request to speak to someone.
When any of these appear, the right outcome is a quick, clean handover: the conversation summary, the account history and the reason for escalation passed to a trained agent, so the borrower does not have to start the story again.
A sensible division of work
| Situation | Who handles it |
|---|---|
| Forgot or paid late; needs a reminder | WhatsApp reminder and payment link |
| Willing, needs a date | Voice agent takes a promise, in the borrower’s language, within contact hours |
| Promise broken once | Re-queued automatically for follow-up |
| Hardship, dispute or restructuring | A trained person, always |
| Late bucket, no contact | Field visit with proof of visit |
Done this way, automation does not make collections colder. It makes room. When reminders and routine promises are handled automatically, agents have time for the calls that actually need a human being.
How we built for it
AutoCollect is designed around this split. Reminders and payment links go out on WhatsApp, voice agents take promises within contact hours and approved scripts, and disputes, hardship and restructuring go to a trained person; they are never automated. Every call, by people and voice agents alike, is summarised and scored, so you can see whether hardship cases were recognised and routed. For the rules side, our collection call compliance checklist is a useful companion, and the broken promise to pay playbook covers the line between won’t-pay and can’t-pay. More in the AutoCollect guides.
Questions
Should AI voice agents handle borrower hardship?
We think not. Automation should detect hardship signals and hand the conversation to a trained person with the summary and account history.
What signals suggest a borrower is in hardship?
Mentions of illness, job loss, bereavement, disaster or business closure; repeated broken promises pointing to inability; distress in the conversation; or a request to speak to someone.
How should collections disputes be handled?
As a question of fact, by a person who can check the account and correct errors, not as an objection for a script to overcome.
See it on your own data. AutoCollect — The right borrower, the right channel. Book a 30-minute working session with an engineer.