Where Should AI Agents Stop and Live Agents Take Over in Collections?
A collections agent handling phone calls might close 50–70 conversations in a day. In a digital collections operation supported by AI and automation, that figure can reach 380–420.
That’s a significant difference. But it doesn’t come from asking agents to work harder. It comes from changing how the work is managed.
For collections leaders facing increasing workloads, recruitment challenges and pressure to control costs, automation offers a way to increase capacity without necessarily increasing headcount. But how much of the collections process should be automated? And where do human agents still add the most value? The answer isn’t necessarily to automate more. It’s to automate the right parts of the customer conversation.
Start with the conversations that don’t need a live agent
Think about how much time collections agents spend on routine tasks. Identity verification, checking balances, answering common questions, confirming payment arrangements and sending reminders are all necessary parts of the collections process. But they don’t always require human involvement. These are often the best places to introduce automation.
Take a customer who gets in touch to explain that they can’t make their full payment this month but could pay half next Tuesday. AI can identify what the customer is asking, capture the proposed payment amount and date, and guide the conversation through the next steps.If the arrangement meets the business’s rules, the system can automatically accept the proposal and schedule a payment reminder.There’s no need for an agent to manage every stage of that interaction. And when you consider how often similar conversations happen across a collections operation, the potential time savings become substantial.
What does this mean for agent productivity?
One of the clearest examples comes from a collections team discussed during our recent webinar on automation and agent efficiency. The team was managing approximately 2,500–3,500 digital conversations a day, with an average of just five agents logged in at any one time. Automation was handling much of the activity, including conversations taking place in the evenings and outside normal working hours.
The difference compared with traditional phone-based collections was considerable.
- 50–70 conversations: The typical number of conversations closed per day by a phone-based collections agent in the example.
- 380–420 conversations: The number closed per agent per day in a digital operation where automation handled more than 75% of the work.
These figures don’t mean individual agents were manually handling hundreds of conversations. They reflect the combined productivity of agents and automation. And that’s an important distinction. Instead of spending time on every routine interaction, agents were available to handle conversations that needed their attention.
For collection teams looking to manage growing volumes without continually adding staff, this is where automation can make a real difference.
But not every collections customer conversation should be automated
Collections conversations aren’t always straightforward. A customer might explain that they’ve lost their job, are struggling financially or are dealing with circumstances that affect their ability to pay. In these situations, a standard automated response may not be appropriate.
AI can help identify potential vulnerability, recognise changes in customer language and highlight when a different approach may be needed.It can also capture information that the customer has already provided, so an agent doesn’t have to ask the same questions again. But there still needs to be a clear decision about when to involve a person.
Some customers will need reassurance, others may need specialist support, and some situations will require judgement that goes beyond an automated process. The aim should be to recognise these situations early and make the handover as straightforward as possible.
That also applies outside normal working hours. If a customer needs an agent who isn’t available, the automated system should explain what happens next and arrange an appropriate follow-up.
Good automation isn’t about keeping customers away from agents. It’s about making sure agents are involved when they can make the biggest difference.
Removing unnecessary steps improves the customer experience
There’s another important consideration when deciding what to automate: how easy is the process for the customer? An automated journey that requires customers to move between different channels, repeat information or navigate complicated menus can create more problems than it solves. One collection’s team featured in the webinar took a relatively simple approach to reducing inbound calls.
The business was sending letters to customers who needed to get in touch. Rather than requiring everyone to call, it introduced QR codes on the letters that opened a WhatsApp conversation. Customers could then ask questions and resolve their enquiries through an automated digital journey.
The business reported more than 90% automated resolution for this part of the process. This reduced the need for inbound calls and freed up agent capacity. It also gave customers a straightforward way to engage without waiting on the phone.
The lesson here is that automation doesn’t always require a complete redesign of the collections operation. Sometimes, changing how customers enter a conversation can deliver significant benefits.
Where should you start?
One of the mistakes made is trying to automate too much, too quickly. It’s understandable. When the potential efficiency gains are significant, there’s a temptation to apply automation across as many processes as possible. But that approach can create poor customer experiences if the journeys haven’t been properly designed.
A more practical starting point is to look at the work agents are doing today. Which enquiries come up most often? Where are agents spending time on repetitive tasks? Which parts of the customer journey create delays or unnecessary handovers? Start with processes that have clear rules and can be handled consistently. Then look at the results.
Are customers completing their journeys? Are agents handling fewer routine enquiries? Are more conversations being resolved without additional intervention? And importantly, are customers who need human support getting it?
As the business gains confidence, automation can be introduced into more parts of the process. This also gives teams the opportunity to adjust journeys based on what customers are actually doing, rather than what they expected them to do.
It’s about making better use of your people
The real opportunity for collections teams isn’t simply to automate as many conversations as possible. It’s to make better use of the resources they already have. When automation handles routine enquiries, agents can spend more time on complex cases, customers experiencing financial difficulties and conversations where their experience and judgement matter.
At the same time, customers benefit from being able to check balances, make payments and resolve straightforward queries at a time that suits them. For collections leaders, the question shouldn’t be whether AI can replace an agent. It should be which parts of the process genuinely need one. Getting that balance right is what makes automation valuable for the business, its agents and its customers.
ON DEMAND WEBINAR
See how collections teams are putting this into practice
In our on-demand webinar, Automation and Agent Efficiency: Doing More with Less in Credit & Collections, we explore real customer examples and operational results, including how collection teams are increasing agent productivity, managing thousands of digital conversations and deciding when automation should hand over to a person.
90 Sec Snippet: Automation and Efficiency Journey
Watch Full Webinar On Demand
"*" indicates required fields