.png)
Every customer's financial situation is different. So why make every communication strategy the same?
In this episode of Break The Cycle we chat with Jacob and Jacopo from our data team about how machine learning can be used to tailor communication strategies to the customer and improve the experience.
Listen on Apple Podcasts:
Transcript
Emma: Hi, welcome back to the Ophelos Mini Series, Break the Cycle. Today we're talking to Jacob and Jacopo about machine learning and how it's revolutionizing the debt industry. So over to you guys, would you like to tell our listeners a little bit about yourselves?
Jacob: I'm Jacob Goss, the Data Lead at Ophelos. So my job is working out how as a company we use data to drive decisions and this comes down to basically how we apply machine learning across all aspects of our business, and I'm here with Jacopo who I'll let introduce himself.
Jacopo: Yeah, I'm Jacopo Attolini a data Analyst at Ophelos and I help Jacob with all that he just mentioned, how we use data to drive our business.
At Ophelos we're trying to improve the collections process for people in debt and businesses where people fall into debt. So what this means for customers is we are trying to find the best ways to encourage customers to pay off their debts in full and give them the best experience and the best resources to help them do that.
Emma: I'd like to get an understanding of what the current status quo is. So businesses blindly send comms, and they're unsure of the impact that they are having on engagement or the people they're actually speaking to?
Jacob: Yeah, so traditionally with debt collection agencies, they are purely focused on just getting payments from customers, and given that their businesses are quite heavily commission-based, this, we think, drives the wrong incentives, and the process is quite fixed and rigid.
So this would typically involve sending maybe an SMS on day three, an email on day five, and this would be the same process for, or the same strategy for each individual customer. So meanwhile, while these communications are going out, you would have agents phone calling customers with the highest balances or the highest propensity to pay just so that they can try to get payments in, so that this benefits them, and what this means for customers and the people in debt is that they're being harassed, they're being bombarded with messages, and it's expensive for the business to run, and it's just a bad experience for the customer.
Jacopo: Yeah, I think Jacob you raised a really important point on the difference that we have with other companies and debt collection agencies. Our kind of machine learning and the use that we have on machine learning allows us to create a communication strategy for each individual customer. So traditional debt collection agencies tend to have one strategy for every customer, and if you look at our communications with customers, they're essentially all different. So each customer is gonna have a different number of SMS that they receive, a different number of emails, and that allows us to tailor our strategies to the individual customer, and we do that using machine learning.
.png)
Emma: So because I naively don't know that much about machine learning, and I'm sure some of our listeners don't, could you give us a high-level overview of what that means and how quality in this case seemingly is more important than quantity?
Jacob: So machine learning is a subset of artificial intelligence and ultimately it comes down to training a model to use data to make decisions. Machine learning has got very powerful over the last, let's say, 10 to 20 years, at making very accurate decisions when given enough training data. This isn't a static process. The model's constantly learning and it works out on a daily basis what is the best thing it can be doing for each of our customers.
Emma: And then I guess is there anything else that you guys wanna share, or any nuggets from machine learning, or from your experience working in the data team at Ophelos?
Jacopo: I guess just in general, I really believe that data has not been used enough in this industry, in the debt industry. We don't know if it's a really old industry with incumbent players that do not allow for innovation, but we really think that there are so many use cases for data that we can use to help our customers be rehabilitated and pay off debt. And this is the reinforcement learning algorithm that we talked about, it's just one of the examples that we can give of how we can help customers using data internally, and that obviously helps the businesses that are gonna choose our services.
Emma: I think with everything going on in the world right now, it's the best time for a change. So you guys are working on some really exciting stuff. It was great having you both, thank you so much. That was really interesting and I've definitely learned a lot.
This is a miniseries brought to you by Ophelos, the Ethical Debt platform. If you want to know more, follow us on LinkedIn or read our latest blog post at content.com. Thanks for listening.
