Online lending is a prime example of an industry tackling the ever-present issue of ease of customer journey vs secure customer identification.
Online applications are a perfect testing ground for cybercriminals to attempt to commit fraud so organizations must ensure their defenses are resilient enough to only approve good customers.
Yet, new sources of data provide exciting new opportunities for both fintech companies and banks to provide a super streamlined process.
Speakers
- Corinne Lleti, Commercial Sales Director, Provenir
- Teresa Byrne, CCO, Dividebuy
- Evgenia Ageykina, Chief Risk Officer Spain, ID Finance
Hosted by
- Gergo Varga, Product Evangelist, SEON
Key takeaways
Building the best risk models & processes – not just the standpoint for regulators
It’s important to look outside of just the risk models because the risk model is only one part of a longer, bigger process as well as how that whole process can build customer delight so that the customer completes the process and get us from A to Z. I think that’s one thing that’s really important, especially when you’re handling online business with a customer, is remaining frictionless as possible.
The second thing that’s really important, I think is that when you have those risk models, as part of that process, they must reflect your ambition as a company.
If you’re targeting a certain sector, the risk models that you put in place have to be able to enable you to go into those new sectors in a way that’s safe and with it, with a level of risks that you can accept.
It’s also looking at how you can combine different verification methods in a way that the customer feels comfortable with the journey. – Corinne Lleti.
—
It’s about trying to identify the minimum. Customer behavior has dramatically changed over the last few years and people just do not want to fill in the great, long application form that’s been put together, for example.
It is much easier to integrate information from another source. Ensuring the continuous improvement of utilizing alternative data sources also helps you to improve customer satisfaction with your service and not seek out other lenders.
We also prevent the people from getting angry since for example some people can feel like its intruding and can be quite suspicious. ‘Why do they ask about this data?’ And then leave the service without proceeding, although some of this data can be publicly accessed without their input. – Evgenia Ageykina.
What are the key challenges when implementing and operationalizing AI?
We operate a semi-autonomous sort of engine where we utilize AI but with human oversight. I think of the credit manager many years ago when I started in this world, it was all very manual. Now I see the credit manager’s job as watching AI understand the machine learning.
I don’t think you can completely remove the human touch because I do think that allows you to use the intelligence that you’ve got to make better decisions. So for us, it’s definitely, that we approach it in a semi-automated way. We allow the AI to do its job but do not rely entirely on it.
I would say we’re interrogating those decisions, at a credit manager’s level, consistently. It’s a continual process. But also, when you talk about AI, we’ve got vulnerable customers as well, that we need to make sure that we’re making the right decision decisions with those consumers.
Segmentation is also really important for us and we spend a huge amount of time on it. It’s key to analyze the initial data and understand the cohorts of consumers in our world, again through a combination of human intelligence and AI that lead to different approval paths. – Teresa Byrne.
—
If we are talking about some traditional markets and traditional products, I would say their basic profile of the customer is quite stable in terms of social-demographic characteristics, in terms of payment behavior, and so on and so forth.
But if we’re talking about online business, the flow can change quite fast. For example, if you change your acquisition channels you’re in for a change with customers as well, and you have to adapt. The more consistent data democracies and factors you have within your system can help you to rework around these new customers.
That goes that as well for different types of products, we need to consider that the profile of the customer can be completely different.
For example, if we’re talking about trends, we’re talking about sub from customers that it is different and they behave not only taking into account credits but there. I would say payment fingerprints would be very different and their preferences as well.
You have to take this into account when you’re segmenting because, for one large segment you can split it, for example, using some formation, then maybe this will not work so well for another segment.
There is a kind of hierarchy of segments, which you have to take into account and which you have to find out by yourself for your particular business through testing and experimentation. – Evgenia Ageykina.
As we enter turbulent times, how do you see the future of online lending?
The importance of having those feedback systems so that we know the mechanisms that worked yesterday is still working today? If so, are they still gonna work tomorrow? Because if we’re not ahead of that game, very easily, it impacts the bottom line.
The speed of change is so much more rapid than it’s ever been before and I think that will continue. If you look at the BNPL industry, it’s quite a new industry. If you look at the uptake of how long it took to get to the same number of people that embrace the credit cards to the same number of banned people, it’s nothing, no comparison whatsoever.
The population is changing and embracing new products, the technology is enabling us to launch new products that were never envisaged before.
I think all of that is just going to advance and get quicker, which means that for people in our world, the challenge gets quicker. – Corinne Lleti.
—
I do think what will change is the way that we do our underwriting today, as we go through this cost of living crisis, we just need to be challenging the data and the decisions continually because a person that looks okay today might not be tomorrow and we need to make sure that we cater to that.
That’s the area but I do think from a technology perspective, this industry will continue to boom, and let’s hope that being able to offset potential harms. – Teresa Byrne.
