Loraine Lawson
Good day. Welcome to the Buzz, a Bank Automation News podcast. I’m Editor Loraine Lawson and recently I spoke with Jeff Keltner, who runs the podcast Leaders in Lending and is the senior vice president of Business Development at the automation lending company Upstart. We talked about the trends in lending, starting with the perils of digitalizing manual processes without automating.Jeff Keltner
You know, probably the most common topic is some nature of digitization. Like, what? And I don’t think just I think everybody, my concern with leaders and lending when we started it was if I bring a bunch of bankers on and ask him what’s going on in lending, they’re all gonna go, deposits are down or up, loan demand is down, and digital is the future. I go, Well, that’s a pretty boring podcast if you keep doing that same message. But I think a lot of what we’re talking through is what makes a good digitization strategy. Where are they focused? There’s been a ton of conversation around blending high touch experiences with highly automated digital experiences, and how do you kind of meld those into the best user experience overall. And a lot of discussion of the need to not just kind of digitize for the sake of digitizing, but to really transform the processes into a better experience. And I think one of the guests from Liberty bank said that if you start with a digital goal and not a process transformation goal, powered by digital, you’re really going to miss the mark. And you’ve got to start with that kind of process reengineering. So that’s there. And then I will say the other episodes people really like are the kind of the really cutting edge of things that are new and different. The the BNPL where’s that gonna go the crypto what’s, what is that? Is it going to be a thing? Is it a is it a Ponzi scheme? Is it a Is it the future of finance? You know, there’s a lot of differing opinions. I think people are really in Interested in what those what those things represent in terms of the future of the industry?

Loraine Lawson
I will say you talked about process and the importance of process engineering. And of course, I think that’s very important with automation. So how can technology, especially automation technology help help them digital as it helped them with this process?

Jeff Keltner
Well, I think it’s, it can help. But I think ultimately, it’s kind of, you’ve got to reorient the goal from digitize. You know, I gave this example yesterday in another conversation, my wife is a physician. And she had an example of digitizing a doctor’s office during COVID. And they like they bought 20 iPads and a nurse walked an iPad for each patient into the patient room, the doctor went room to room talking to different patients on different iPads over zoom, but like they didn’t like they were in the physical exam rooms, he was walking around to all 20 rooms one at a time. And they were setting up in each room. And I think it’s a funny example. And I give them a lot of credit, that was a very short term thing. But I think it’s it indicates to you, when we think of digitizing as the goal, you end up doing that you digitize the old way of doing things. And what people really want is an easier experience enabled by digital. And so digital isn’t the goal, it’s the tool. And the tool needs to be easier process. And that’s where automation really comes in. So things like, you know, if you’re trying to do ID verification, you know, in the branch, you just ask for an ID. And that was easy. And if we had it because I can I got my wallet, I just pull it out hands it easy. Now we say well, the digital version is take a picture of the ID. And that’s like a way to do it. But there’s also a lot of third party tools, automated systems that can do checks. A lot of times even if you take the picture, the person looks at it as opposed to some of the software that can do automatic automatic recognition of valid IDs. And are you looking at other signals that could give you confidence in the identity of the person? So to me, it’s really that, you know, if I want to look at bank statements, can I integrate with the bank, depository my system or use plaid or Yodlee, or with other systems to pull in data and look at that, as opposed to saying, Please upload your last four months of bank statements, then my underwriter will look at all of them and try and determine if your income is represented there. As you stated it, I think that’s like, that’s what happens when you have digital as the goal of digitize it, right, you uploaded the document from your phone, but you still waited two days for the person to review and get back to you as opposed to did I automate it and make that process easier for the consumer. So to me, that’s, that’s the goal really is the easiest process for the consumer digital just gives you so many more tools, and so many ways to automate what used to be manual processes. And I think that’s got to be the true north. Otherwise, you’re gonna end up with that kind of doctor walking room to room, or, you know, the lender I talked to who digitize their auto process for auto loan, but you still had to walk into the branch to sign the paperwork and get the check. And I went well, like, it’s great that I could apply for my phone. But if I got to come into the branch at the end, you didn’t really make it that much better for me, you know, you kind of you missed the mark a little bit on a really digital experience.Loraine Lawson
Yes, it’s a paradigm shift in a way right to think about automation and, and automating rather than just duplicating through technology. So what do you think is missing there in our ability to make that shift? Is it just a lack of knowledge of solutions? Or what what do you think causes that?Jeff Keltner
I think it’s, it’s a number of things, lack of knowledge of solutions is one, I think you just have, particularly in financial institutions, so many stakeholders, right, that have a perspective. So the risk team is not comfortable with, you know, they’re comfortable with the way you did ID verification of the branch, and they just want to like copy and paste and liftover. And so I think too often we don’t get the whole institution, all the stakeholders from credit and risk and compliance, along with product involved in the disc discussion, bring them in early and really have a conversation about how do we meet our requirements in a new and different way. And that’s, it’s the path of least resistance, right? It’s just like, I don’t want to have to go mess with compliance and risk when I’m building this new thing. So I’ll just do the thing that they’re already comfortable with. And so I think it’s, it’s that and it’s just, it’s a new muscle. That’s it’s been the way we’ve worked this is this is kind of unique moment, over the last decades when you can truly transform a process and it’s just not a muscle. We as institutions have is like totally reengineering a process. We’ve shifted them, but we haven’t really totally transformed them in this way. And so I think it’s just, it’s, it’s making sure you get the right people in the room early that you’re asking the right question, right? And you’re really thinking about what you’re trying to do. And, and being willing to go out look and rethink about how you do things and think about things in a different way. And sometimes it’s just, you know, it takes a it takes a village so to speak. I mean in a bank or credit union there are a lot of people have to get on board with that plan. It’s easy for any one of them to stand in the way and I think You know, getting that, that that team, you know, all rowing the same direction and seeing the same vision is really the key thing that you have to do this. It’s hard. It’s not easy. And it’s not the natural, I think inclination for most people in the industry.

Loraine Lawson
It takes a village, but it strikes me It may also take an executive sponsor. I mean, and I think about this in terms more of citizen Developer Programs, I mean, I wonder if that creates a risk of automating what we have, rather than automating what we could do?

Jeff Keltner
Yeah, I think you’re right, it takes an executive sponsor and the executive sponsors I’ve seen, but the most effective are those, I think there’s a, an inclination to as a business owner, say, Hey, this is what I want to do. And then you like, you do all your due diligence, you have your whole plan, that you bring it to your legal team, your compliance team, your risk team, your credit team, like fully baked, right, and you go, here’s what we’re gonna do. And then of course, the National Enquirer, developers go, Oh, my God, like, you’re totally doing everything, not the way I want. And the best executive sponsors, I see, see the kernel of a great idea, it could come from a citizen developer, or anywhere else, right? That it’s an idea, they go, Hey, let’s go get everybody in the room and talk about how we can make the process better, and make it better for the institution make it better for the customer. And let’s get a plan together. And that’s what I really mean by it takes a village, if you wait till the very end, to get a stamp of approval from all of the committees, they’re all going to react negatively, because it’s like a change in the way they’re thinking and they weren’t involved in the design. And they may come and go, Hey, I liked this automation, I automated ID verification, but I need these two things, or this fallback process. If these things are high risk, what are we going to do, and they get comfortable with the plan. And I think getting all those stakeholders involved early in defining the solution is really critical. And if you don’t do that, then you’re just gonna like, I had this brilliant idea. And I ran into a brick wall, and I couldn’t get it over the line. Right. And we’ve seen that too. And I think that’s the process to get people involved early. And it does take to convene that those groups, right, it takes a relatively senior person often in the bank. So you’ve got to have a sponsor, that’s going to say, hey, I want to bring these constituents together, because I think there’s value in whatever that project digitization effort is. And that’s, I think, critical to getting it done.

Loraine Lawson
You have a unique position, not just as a podcast host, your Senior Vice President of Business Development at Upstart, which is this hugely successful startup. I wondered what opportunities either you or Upstart in general sees for automation or technology that you think most institutions are missing right now that they’re just totally, like, maybe it’s an example you’ve seen or?

Jeff Keltner
Yeah, I just think, I don’t know that they’re missing it. But it’s the magic is in the execution, and the willingness to go do it. And I’ll give you give examples. And I won’t name my bank, because it’s, I don’t want to call anybody out. But like, I went through a mortgage refi process, right. And it was fully digitized, right. But it was still loan officer centric. And there was still like, Please upload this document, please. And all of these were things that were API enabled. And they just kind of they built the old process, they you know, and there’s, I get in that case, there’s like a whole bunch of loan officers who want to get paid. And if I went through a self serve process, like, that’s a big institutional change, that’s a big problem. But I think anytime you see that human interaction in a transaction, I’ll say, in an administrative role, that’s an opportunity for automation. And the interesting thing, we talked a little bit about this kind of balance of automation and high touch human. And what I see the best institutions doing well is saying, when someone can self serve, they should, and we should make it super easy. I don’t want to make you get on the phone. You know, I had this was buying a car, I like configured the car I wanted. I said, I’d like to order this. And I said, please call us. I said, Well, why don’t I want to call you I just like, and then I got on the phone, they say can you walk me through the configuration you want? I’ve got the app open right, here I go. I did that online. Just like send me the car. And I Well, yeah, come into the branch to confirm I don’t want to come into the dealership, I just want to like, I want to buy the car. And so we often put the human in the middle of a customer who just wants to transact, right, we know that like nobody wants to call to check their balance. They want to like, pick up their phone see it? I think often we do that. And the thing that I think is people sometimes mistake because they think that means you want to do it everywhere. And I don’t I think you want to take your people and put them at the point where the customer needs them. Right? Hey, do I What’s the difference between this 15 year and 30 year mortgage? And I see this interest only or this arm like what are these things? And how do I make a smart choice? I want to personally what’s a Roth IRA and an IRA and a 401 K and like, what’s a money market account and a CD and a savings account? Which of these should I be using and what? Like I don’t most consumers need help navigating those choices. Now once I’ve said hey, I need a Roth IRA. I don’t need someone to like have requiring me to get on the phone to make a deposit or to open the account. I want to automate that transactional administrative process as much as possible. But I want to make sure that those people are available and easy to get to Don’t make me call in and go through seven levels of phone tree. Because your staffs all on the phone, verifying identity and asking for documentation like free your staff up to be immediately available to help people in that moment. And so I think that because I’m sensitive to people who want to automate, what to automate everything, I think is a misnomer. But we just there’s so much I mean, you go through any lending process today, any account opening process, and there’s almost always a, call us upload something and whenever you’re uploading somebody is on the other end, looking at the thing that you uploaded. And that’s almost always something that could be done either with, you know, visual recognition on an uploaded document, or often, you know, we do this in auto refi, a lot of times you say, give me the VIN on your car? Well, it turns out, that’s a really high friction thing for a customer to do go get the VIN on their car, if you just give me your name and address almost all the DMVs have a database where I can kind of like a credit poll, figure out the VIN on the vehicle, and I can connect it to the loan that you have. And so we can take that would have been frictional process of Give me your VIN and just say, hey, we found we found this, you know, 2010 Toyota Corolla is that the car you’re trying to refinance, and it looks like you got a loan at, you know, Bank of America is that the loan and like, and you can connect those things very, very clearly through automation. So I just think you’ve got to really think about how do I do that and take the burden off my people to do frankly, boring administrative tasks and off the consumer to do a bunch of work in order to get to a place that you could have done without making either of them do those things.

Loraine Lawson
Yeah, I think institutions maybe are struggling with that where to automate versus where not to automate question, I think that’ll be a strategic differentiator for the future. Yeah, I agree. So upstart, you, you use factors beyond credit score to determine credit work rather than this, like education, employment. I know, auto loans are a huge use case for upstarts facilitated 1.5 billion and auto loans on its platform in 2022. But I wondered where else?

Jeff Keltner
Yeah, that’s the goal. I mean, 2020 is not over. So that’s what they tell

Loraine Lawson
you one. Sorry. I misspoke. My apologies. But where else? Are financial institutions applying your solution? And are there any surprises for you?

Jeff Keltner
Oh, there’s always surprises. But um, you know, we started so our underwriting model really began in unsecured personal loans. And that’s still the bulk of the originations we do with our lending partners. And, and that’s for a very specific reason, which is, if you think about what you said, we use credit scores, data points beyond just a credit score to help understand the risk of a given borrower for a given loan, and help you then approve the fundamental understanding we had that we came to that, you know, why do you leave Google and start a FinTech company? I mean, what what we saw on the market was that many more Americans are credit worthy than have credit scores that would cause a typical financial institution to lend to them, right. And I think everybody in finance knows this to be true. And maybe this is the thing we learned is how under appreciated this fact, is, because you look at like a subprime loan pool, that a bank would go, I could never be there. And it’s 20, or 25%, losses or 30% losses, right. And that, that still means it’s 70, or 75, or 80%, good borrowers. And we’re turning away the 80%. Because we couldn’t identify the 20. into me, that’s the surprise of the learning is like, if you think of your goal in credit underwriting as managing to a particular loss target, you can do that with credit scores and other things. I have a 3% loss portfolio, but I do it at the expense of declining a lot of applicants who would have paid me back, right? That whole subprime pool that I just said no to all of them with a low credit score, turns out 80% of them were gonna pay me and I declined all of them. And I think that was underappreciated in the industry, what we came in to do. And the reason I say that is the place where that’s most true, where that most impacts the ability of the consumer to get a loan isn’t an unsecured loan, because there’s no, there’s no collateral to say, Hey, I’m okay with a slightly lower credit score, because at least I’ve got a lien on the car, or I’ve got a lien on the home, like don’t have that. And so we’ve kind of worked our way from the place where that understanding of payment risk is the most important to expanding access to credit and worked our way you know, unsecured and we’re working our way down, down the path towards auto and you can imagine going towards mortgage and small medium business. So we’re primarily today in auto and personal loan, we’re actively working in small medium business, smaller dollar loans, something more akin to what people use payday for. So shorter duration, smaller dollar, the average person was like 10,000. What about somebody needs 500 or 1000? We’re working on that and looking at mortgage as well because I think the fundamental fact that more Americans are low risk than we understand to be low risk because of their credit score is true across the board. And it’s just a question of where’s that most valuable? And where do we want to apply that that realization first?

Loraine Lawson
Think pretty open to exploring that with you are the big gain market? Or do you find that? Maybe there’s different range banks that are more interested or less interested?

Jeff Keltner
Yeah, there’s, there’s not like a size or something else that indicates the more or less, I think most institutions I talked to acknowledge that like, I mean, you can look at the data and go, yeah, there’s like, clearly, there’s a lot of people who are credit worthy that we don’t lend to. I mean, every bank knows, ever, every product owner I talk to is trying to go, how do I get that, you know, start approving this next tranche of customers to grow my portfolio to serve my customers better? So I think everybody acknowledges that, that there’s a there’s a mismatch between scoring and real risk, and that there’s probably an ability to close it. So they’re all interested from that level of yeah, this is a real problem, and it should be solvable. Okay. And then there’s a question of like, Am I comfortable with the way you’re solving it? And I think there you see a pretty wide range of people who go, we just can’t, you know, we don’t get it. We don’t want to go there. And a lot of institutions and an increasing number, that are very comfortable with with the approach, and that are saying, Hey, this is this is the future. And we want to be there first. So we can do a better job serving our customers of winning customers in the marketplace by understanding who deserves a good rate and is getting a bad one. And we can go when that customer to our institution. And I think the probably the biggest question is less are the models, right? Like, are you predicting risk accurately? We got lots of data to show that and more a question of, will our regulators be okay, with the way this is being done. And that’s why we, you know, we went as a as a FinTech company, to the CFPB. Very early in our existence, before we had worked with a bank partner that originated single loan, and said, Hey, we’ve got this technology, and we believe it can increase access to credit and decrease the cost of credit. That’s good. But there’s these questions around regulatory compliance, how do we think about fair lending in the context of these variables and this kind of model, and we work with that bit with the Bureau to develop a kind of a compliance and testing program. And that ultimately resulted in their issuing us and no action letter was a three year letter, at the end of three years, they issued us a new three year letter, and we’re kind of in the middle of the application, that letter we report, right, they had a regulator, the bureau, so and that tells me they were interested in engaging, they were great partners at the Bureau. And I think points to you that while individual institutions may be concerned about the regulator, the regulators, we talked to in the central offices, are very interested in enabling these kinds of technologies to open up access and decrease the cost of credit. And, you know, they want to make sure it’s tested properly, that it’s managed properly. But there’s a real willingness to do that. And the banks we’ve seen engaged with the regulators have had a really positive experience engaging on that. So I think it’s a it’s an understandable concern. And of course, the one of the sad learnings, one of the things for me is like, every relationship between an examiner and an financial institution is different, right? It’s not like, oh, the OCC said this like, well, one person at the OCC said that when I walk into a different regional office, or a different examiner and their relationship and their history with a particular institution and how that’s gone, they all vary. And so it’s not a uniform, simple answer. That’s something we learned along the way. But I think there there’s a motion towards a movement towards how do we make this thing possible, because the benefits to both the financial institutions in terms of safety and soundness truly understanding the credit risk on their books, and to the consumer population about actually being able to open up access to traditionally underserved communities is so great.

Loraine Lawson
You’ve been listening to the Buzz, a Bank Automation News podcast. Thank you for your time, and be sure to visit us at Bank automation news.com for more automation news. You can also follow us on Twitter and LinkedIn. Please don’t hesitate to rate this podcast on your podcast platform of choice.

Loraine Lawson
Good day. Welcome to the Buzz, a Bank Automation News podcast. I’m Editor Loraine Lawson and recently I spoke with Jeff Keltner, who runs the podcast Leaders in Lending and is the senior vice president of Business Development at the automation lending company Upstart. We talked about the trends in lending, starting with the perils of digitalizing manual processes without automating.Jeff Keltner
You know, probably the most common topic is some nature of digitization. Like, what? And I don’t think just I think everybody, my concern with leaders and lending when we started it was if I bring a bunch of bankers on and ask him what’s going on in lending, they’re all gonna go, deposits are down or up, loan demand is down, and digital is the future. I go, Well, that’s a pretty boring podcast if you keep doing that same message. But I think a lot of what we’re talking through is what makes a good digitization strategy. Where are they focused? There’s been a ton of conversation around blending high touch experiences with highly automated digital experiences, and how do you kind of meld those into the best user experience overall. And a lot of discussion of the need to not just kind of digitize for the sake of digitizing, but to really transform the processes into a better experience. And I think one of the guests from Liberty bank said that if you start with a digital goal and not a process transformation goal, powered by digital, you’re really going to miss the mark. And you’ve got to start with that kind of process reengineering. So that’s there. And then I will say the other episodes people really like are the kind of the really cutting edge of things that are new and different. The the BNPL where’s that gonna go the crypto what’s, what is that? Is it going to be a thing? Is it a is it a Ponzi scheme? Is it a Is it the future of finance? You know, there’s a lot of differing opinions. I think people are really in Interested in what those what those things represent in terms of the future of the industry?

Loraine Lawson
I will say you talked about process and the importance of process engineering. And of course, I think that’s very important with automation. So how can technology, especially automation technology help help them digital as it helped them with this process?

Jeff Keltner
Well, I think it’s, it can help. But I think ultimately, it’s kind of, you’ve got to reorient the goal from digitize. You know, I gave this example yesterday in another conversation, my wife is a physician. And she had an example of digitizing a doctor’s office during COVID. And they like they bought 20 iPads and a nurse walked an iPad for each patient into the patient room, the doctor went room to room talking to different patients on different iPads over zoom, but like they didn’t like they were in the physical exam rooms, he was walking around to all 20 rooms one at a time. And they were setting up in each room. And I think it’s a funny example. And I give them a lot of credit, that was a very short term thing. But I think it’s it indicates to you, when we think of digitizing as the goal, you end up doing that you digitize the old way of doing things. And what people really want is an easier experience enabled by digital. And so digital isn’t the goal, it’s the tool. And the tool needs to be easier process. And that’s where automation really comes in. So things like, you know, if you’re trying to do ID verification, you know, in the branch, you just ask for an ID. And that was easy. And if we had it because I can I got my wallet, I just pull it out hands it easy. Now we say well, the digital version is take a picture of the ID. And that’s like a way to do it. But there’s also a lot of third party tools, automated systems that can do checks. A lot of times even if you take the picture, the person looks at it as opposed to some of the software that can do automatic automatic recognition of valid IDs. And are you looking at other signals that could give you confidence in the identity of the person? So to me, it’s really that, you know, if I want to look at bank statements, can I integrate with the bank, depository my system or use plaid or Yodlee, or with other systems to pull in data and look at that, as opposed to saying, Please upload your last four months of bank statements, then my underwriter will look at all of them and try and determine if your income is represented there. As you stated it, I think that’s like, that’s what happens when you have digital as the goal of digitize it, right, you uploaded the document from your phone, but you still waited two days for the person to review and get back to you as opposed to did I automate it and make that process easier for the consumer. So to me, that’s, that’s the goal really is the easiest process for the consumer digital just gives you so many more tools, and so many ways to automate what used to be manual processes. And I think that’s got to be the true north. Otherwise, you’re gonna end up with that kind of doctor walking room to room, or, you know, the lender I talked to who digitize their auto process for auto loan, but you still had to walk into the branch to sign the paperwork and get the check. And I went well, like, it’s great that I could apply for my phone. But if I got to come into the branch at the end, you didn’t really make it that much better for me, you know, you kind of you missed the mark a little bit on a really digital experience.Loraine Lawson
Yes, it’s a paradigm shift in a way right to think about automation and, and automating rather than just duplicating through technology. So what do you think is missing there in our ability to make that shift? Is it just a lack of knowledge of solutions? Or what what do you think causes that?Jeff Keltner
I think it’s, it’s a number of things, lack of knowledge of solutions is one, I think you just have, particularly in financial institutions, so many stakeholders, right, that have a perspective. So the risk team is not comfortable with, you know, they’re comfortable with the way you did ID verification of the branch, and they just want to like copy and paste and liftover. And so I think too often we don’t get the whole institution, all the stakeholders from credit and risk and compliance, along with product involved in the disc discussion, bring them in early and really have a conversation about how do we meet our requirements in a new and different way. And that’s, it’s the path of least resistance, right? It’s just like, I don’t want to have to go mess with compliance and risk when I’m building this new thing. So I’ll just do the thing that they’re already comfortable with. And so I think it’s, it’s that and it’s just, it’s a new muscle. That’s it’s been the way we’ve worked this is this is kind of unique moment, over the last decades when you can truly transform a process and it’s just not a muscle. We as institutions have is like totally reengineering a process. We’ve shifted them, but we haven’t really totally transformed them in this way. And so I think it’s just, it’s, it’s making sure you get the right people in the room early that you’re asking the right question, right? And you’re really thinking about what you’re trying to do. And, and being willing to go out look and rethink about how you do things and think about things in a different way. And sometimes it’s just, you know, it takes a it takes a village so to speak. I mean in a bank or credit union there are a lot of people have to get on board with that plan. It’s easy for any one of them to stand in the way and I think You know, getting that, that that team, you know, all rowing the same direction and seeing the same vision is really the key thing that you have to do this. It’s hard. It’s not easy. And it’s not the natural, I think inclination for most people in the industry.

Loraine Lawson
It takes a village, but it strikes me It may also take an executive sponsor. I mean, and I think about this in terms more of citizen Developer Programs, I mean, I wonder if that creates a risk of automating what we have, rather than automating what we could do?

Jeff Keltner
Yeah, I think you’re right, it takes an executive sponsor and the executive sponsors I’ve seen, but the most effective are those, I think there’s a, an inclination to as a business owner, say, Hey, this is what I want to do. And then you like, you do all your due diligence, you have your whole plan, that you bring it to your legal team, your compliance team, your risk team, your credit team, like fully baked, right, and you go, here’s what we’re gonna do. And then of course, the National Enquirer, developers go, Oh, my God, like, you’re totally doing everything, not the way I want. And the best executive sponsors, I see, see the kernel of a great idea, it could come from a citizen developer, or anywhere else, right? That it’s an idea, they go, Hey, let’s go get everybody in the room and talk about how we can make the process better, and make it better for the institution make it better for the customer. And let’s get a plan together. And that’s what I really mean by it takes a village, if you wait till the very end, to get a stamp of approval from all of the committees, they’re all going to react negatively, because it’s like a change in the way they’re thinking and they weren’t involved in the design. And they may come and go, Hey, I liked this automation, I automated ID verification, but I need these two things, or this fallback process. If these things are high risk, what are we going to do, and they get comfortable with the plan. And I think getting all those stakeholders involved early in defining the solution is really critical. And if you don’t do that, then you’re just gonna like, I had this brilliant idea. And I ran into a brick wall, and I couldn’t get it over the line. Right. And we’ve seen that too. And I think that’s the process to get people involved early. And it does take to convene that those groups, right, it takes a relatively senior person often in the bank. So you’ve got to have a sponsor, that’s going to say, hey, I want to bring these constituents together, because I think there’s value in whatever that project digitization effort is. And that’s, I think, critical to getting it done.

Loraine Lawson
You have a unique position, not just as a podcast host, your Senior Vice President of Business Development at Upstart, which is this hugely successful startup. I wondered what opportunities either you or Upstart in general sees for automation or technology that you think most institutions are missing right now that they’re just totally, like, maybe it’s an example you’ve seen or?

Jeff Keltner
Yeah, I just think, I don’t know that they’re missing it. But it’s the magic is in the execution, and the willingness to go do it. And I’ll give you give examples. And I won’t name my bank, because it’s, I don’t want to call anybody out. But like, I went through a mortgage refi process, right. And it was fully digitized, right. But it was still loan officer centric. And there was still like, Please upload this document, please. And all of these were things that were API enabled. And they just kind of they built the old process, they you know, and there’s, I get in that case, there’s like a whole bunch of loan officers who want to get paid. And if I went through a self serve process, like, that’s a big institutional change, that’s a big problem. But I think anytime you see that human interaction in a transaction, I’ll say, in an administrative role, that’s an opportunity for automation. And the interesting thing, we talked a little bit about this kind of balance of automation and high touch human. And what I see the best institutions doing well is saying, when someone can self serve, they should, and we should make it super easy. I don’t want to make you get on the phone. You know, I had this was buying a car, I like configured the car I wanted. I said, I’d like to order this. And I said, please call us. I said, Well, why don’t I want to call you I just like, and then I got on the phone, they say can you walk me through the configuration you want? I’ve got the app open right, here I go. I did that online. Just like send me the car. And I Well, yeah, come into the branch to confirm I don’t want to come into the dealership, I just want to like, I want to buy the car. And so we often put the human in the middle of a customer who just wants to transact, right, we know that like nobody wants to call to check their balance. They want to like, pick up their phone see it? I think often we do that. And the thing that I think is people sometimes mistake because they think that means you want to do it everywhere. And I don’t I think you want to take your people and put them at the point where the customer needs them. Right? Hey, do I What’s the difference between this 15 year and 30 year mortgage? And I see this interest only or this arm like what are these things? And how do I make a smart choice? I want to personally what’s a Roth IRA and an IRA and a 401 K and like, what’s a money market account and a CD and a savings account? Which of these should I be using and what? Like I don’t most consumers need help navigating those choices. Now once I’ve said hey, I need a Roth IRA. I don’t need someone to like have requiring me to get on the phone to make a deposit or to open the account. I want to automate that transactional administrative process as much as possible. But I want to make sure that those people are available and easy to get to Don’t make me call in and go through seven levels of phone tree. Because your staffs all on the phone, verifying identity and asking for documentation like free your staff up to be immediately available to help people in that moment. And so I think that because I’m sensitive to people who want to automate, what to automate everything, I think is a misnomer. But we just there’s so much I mean, you go through any lending process today, any account opening process, and there’s almost always a, call us upload something and whenever you’re uploading somebody is on the other end, looking at the thing that you uploaded. And that’s almost always something that could be done either with, you know, visual recognition on an uploaded document, or often, you know, we do this in auto refi, a lot of times you say, give me the VIN on your car? Well, it turns out, that’s a really high friction thing for a customer to do go get the VIN on their car, if you just give me your name and address almost all the DMVs have a database where I can kind of like a credit poll, figure out the VIN on the vehicle, and I can connect it to the loan that you have. And so we can take that would have been frictional process of Give me your VIN and just say, hey, we found we found this, you know, 2010 Toyota Corolla is that the car you’re trying to refinance, and it looks like you got a loan at, you know, Bank of America is that the loan and like, and you can connect those things very, very clearly through automation. So I just think you’ve got to really think about how do I do that and take the burden off my people to do frankly, boring administrative tasks and off the consumer to do a bunch of work in order to get to a place that you could have done without making either of them do those things.

Loraine Lawson
Yeah, I think institutions maybe are struggling with that where to automate versus where not to automate question, I think that’ll be a strategic differentiator for the future. Yeah, I agree. So upstart, you, you use factors beyond credit score to determine credit work rather than this, like education, employment. I know, auto loans are a huge use case for upstarts facilitated 1.5 billion and auto loans on its platform in 2022. But I wondered where else?

Jeff Keltner
Yeah, that’s the goal. I mean, 2020 is not over. So that’s what they tell

Loraine Lawson
you one. Sorry. I misspoke. My apologies. But where else? Are financial institutions applying your solution? And are there any surprises for you?

Jeff Keltner
Oh, there’s always surprises. But um, you know, we started so our underwriting model really began in unsecured personal loans. And that’s still the bulk of the originations we do with our lending partners. And, and that’s for a very specific reason, which is, if you think about what you said, we use credit scores, data points beyond just a credit score to help understand the risk of a given borrower for a given loan, and help you then approve the fundamental understanding we had that we came to that, you know, why do you leave Google and start a FinTech company? I mean, what what we saw on the market was that many more Americans are credit worthy than have credit scores that would cause a typical financial institution to lend to them, right. And I think everybody in finance knows this to be true. And maybe this is the thing we learned is how under appreciated this fact, is, because you look at like a subprime loan pool, that a bank would go, I could never be there. And it’s 20, or 25%, losses or 30% losses, right. And that, that still means it’s 70, or 75, or 80%, good borrowers. And we’re turning away the 80%. Because we couldn’t identify the 20. into me, that’s the surprise of the learning is like, if you think of your goal in credit underwriting as managing to a particular loss target, you can do that with credit scores and other things. I have a 3% loss portfolio, but I do it at the expense of declining a lot of applicants who would have paid me back, right? That whole subprime pool that I just said no to all of them with a low credit score, turns out 80% of them were gonna pay me and I declined all of them. And I think that was underappreciated in the industry, what we came in to do. And the reason I say that is the place where that’s most true, where that most impacts the ability of the consumer to get a loan isn’t an unsecured loan, because there’s no, there’s no collateral to say, Hey, I’m okay with a slightly lower credit score, because at least I’ve got a lien on the car, or I’ve got a lien on the home, like don’t have that. And so we’ve kind of worked our way from the place where that understanding of payment risk is the most important to expanding access to credit and worked our way you know, unsecured and we’re working our way down, down the path towards auto and you can imagine going towards mortgage and small medium business. So we’re primarily today in auto and personal loan, we’re actively working in small medium business, smaller dollar loans, something more akin to what people use payday for. So shorter duration, smaller dollar, the average person was like 10,000. What about somebody needs 500 or 1000? We’re working on that and looking at mortgage as well because I think the fundamental fact that more Americans are low risk than we understand to be low risk because of their credit score is true across the board. And it’s just a question of where’s that most valuable? And where do we want to apply that that realization first?

Loraine Lawson
Think pretty open to exploring that with you are the big gain market? Or do you find that? Maybe there’s different range banks that are more interested or less interested?

Jeff Keltner
Yeah, there’s, there’s not like a size or something else that indicates the more or less, I think most institutions I talked to acknowledge that like, I mean, you can look at the data and go, yeah, there’s like, clearly, there’s a lot of people who are credit worthy that we don’t lend to. I mean, every bank knows, ever, every product owner I talk to is trying to go, how do I get that, you know, start approving this next tranche of customers to grow my portfolio to serve my customers better? So I think everybody acknowledges that, that there’s a there’s a mismatch between scoring and real risk, and that there’s probably an ability to close it. So they’re all interested from that level of yeah, this is a real problem, and it should be solvable. Okay. And then there’s a question of like, Am I comfortable with the way you’re solving it? And I think there you see a pretty wide range of people who go, we just can’t, you know, we don’t get it. We don’t want to go there. And a lot of institutions and an increasing number, that are very comfortable with with the approach, and that are saying, Hey, this is this is the future. And we want to be there first. So we can do a better job serving our customers of winning customers in the marketplace by understanding who deserves a good rate and is getting a bad one. And we can go when that customer to our institution. And I think the probably the biggest question is less are the models, right? Like, are you predicting risk accurately? We got lots of data to show that and more a question of, will our regulators be okay, with the way this is being done. And that’s why we, you know, we went as a as a FinTech company, to the CFPB. Very early in our existence, before we had worked with a bank partner that originated single loan, and said, Hey, we’ve got this technology, and we believe it can increase access to credit and decrease the cost of credit. That’s good. But there’s these questions around regulatory compliance, how do we think about fair lending in the context of these variables and this kind of model, and we work with that bit with the Bureau to develop a kind of a compliance and testing program. And that ultimately resulted in their issuing us and no action letter was a three year letter, at the end of three years, they issued us a new three year letter, and we’re kind of in the middle of the application, that letter we report, right, they had a regulator, the bureau, so and that tells me they were interested in engaging, they were great partners at the Bureau. And I think points to you that while individual institutions may be concerned about the regulator, the regulators, we talked to in the central offices, are very interested in enabling these kinds of technologies to open up access and decrease the cost of credit. And, you know, they want to make sure it’s tested properly, that it’s managed properly. But there’s a real willingness to do that. And the banks we’ve seen engaged with the regulators have had a really positive experience engaging on that. So I think it’s a it’s an understandable concern. And of course, the one of the sad learnings, one of the things for me is like, every relationship between an examiner and an financial institution is different, right? It’s not like, oh, the OCC said this like, well, one person at the OCC said that when I walk into a different regional office, or a different examiner and their relationship and their history with a particular institution and how that’s gone, they all vary. And so it’s not a uniform, simple answer. That’s something we learned along the way. But I think there there’s a motion towards a movement towards how do we make this thing possible, because the benefits to both the financial institutions in terms of safety and soundness truly understanding the credit risk on their books, and to the consumer population about actually being able to open up access to traditionally underserved communities is so great.

Loraine Lawson
You’ve been listening to the Buzz, a Bank Automation News podcast. Thank you for your time, and be sure to visit us at Bank automation news.com for more automation news. You can also follow us on Twitter and LinkedIn. Please don’t hesitate to rate this podcast on your podcast platform of choice.