Jeff Keltner:
I was asked when I applied for a mortgage, "Please provide three years of tax returns to validate your income." And I said, "You've been my bank for 20 years. I've direct deposited every paycheck into my deposit account. You shouldn't ask me that."
James Robert Lay:
Greetings and hello, I am James Robert Lay, and welcome to Episode 251 of the Banking on Digital Growth Podcast. Today's episode is part of the Exponential Insight Series, and I'm excited to welcome Jeff Keltner to the show. Jeff is the Senior Vice President of Business Development at Upstart, who is on a mission to enable effortless credit based on true risk.
Jeff is also the host of the Leaders in Lending Podcast, and today, I look forward to talking with Jeff about the opportunities for financial brands to create and capture when it comes to lending and AI. Welcome to the show, Jeff. It is good to share time with you today, buddy.
Jeff Keltner:
Thanks for having me. It's great to be here.
James Robert Lay:
Before we get into talking about the trends and the transformations happening around the consumer lending landscape, what's good for you right now? Personal, professional, it's your pick to get started.
Jeff Keltner:
I mean, personal's always I got kids, family. I was at my college reunion, I won't give the year number so I won't age myself, but somebody asked, "How's life?" And somebody else responded, "Work and kids, that's the whole story." But when the family's happy and healthy and everybody's doing well, that's about all you can ask for, I think.
James Robert Lay:
I'm right there with you. My wife and I, we have four kids that are 12, 10, 8, and 6, and just coming out of the month of October, oh my goodness. October was I think the busiest month that we have had since probably 2019, considering the world slowed down for a couple of years, but it was a wild ride in October, but it was a good one.
Jeff Keltner:
I've got 13 and going to be 11 this week.
James Robert Lay:
Oh, you're right there.
Jeff Keltner:
I'm right there with you. Not as many. I always say when you have to go from man to man to zone, then you've got a challenge when you get outnumbered, so that's a different beast, but-
James Robert Lay:
We have been dividing and conquering, my wife and I, because especially for the 12-year-old and the 10-year-old, and even the eight-year-old, their activities in school, sports, it feels like an exponential increase. I think the good news is we're unified in working together to make the best of everything when we can. I think the idea of complexity too is one, if we can parlay from the home front and kids to the world of consumer lending, money is complex, lending is complex.
People need credit, people need loans. But I want to look at this from a human standpoint first. Even in a digital world, digital lending, people still need the human touch and you note that this is a problem that most lenders fail to understand. Why is this? Why do people still need people when it comes to lending in a digital world?
Jeff Keltner:
I mean, I think sometimes they do and sometimes they don't. I think this is one of the hard balances for any... whether it's lending or wealth management or just investment management, any experience is you want to be able to self-service sometimes. And then when you need a human, usually I find that that moment is when you don't understand something or something's gone wrong, I've got a challenge.
We see that a lot during COVID, and now as the economy is a little bit more challenging for many consumers, they say, "Hey, somethings gone wrong. Can you help me with a payment plan, a forbearance, a hardship?" And I think the reality is you to don't want to be forced to talk to somebody when you're ready to self-serve and you know what you want. And you don't want to be forced to try and figure it out on your own when you're not sure what you want or you need help figuring it out.
And the most successful companies make it easy to avoid talking to a human when you don't want to, and super easy to talk to a human when you do. And sometimes, we make it the opposite. We make you talk to somebody when you don't want to and we make it really hard to get ahold of the person when you need them. And I think that balance is really hard to achieve, but it is really important.
James Robert Lay:
It's two sides of the coin. On one side, it's the complexity. I've reached a problem, I need help. The only way that I can get help outside of that self-serve and resolve that conflict is by gaining clarity through a conversation or a connection, if you will.
Doesn't have to necessarily be a conversation like we're having, could be chat, et cetera, with another human being. Where and when? I think that's a great question. Where and when should financial brands focus on human to human interaction, and when should they step away and leverage the technology, leverage the self-serve, back to your point?
Jeff Keltner:
I think it's an interesting question, but I think it's the front and the back, where they want to be in the middle of it, in the middle where they don't. So let me maybe describe that in a little bit more depth. I'm having my air conditioning replaced, do I need a personal loan or should I do a home equity line of credit? I'm being told about this cash out refi.
James Robert Lay:
Wow.
Jeff Keltner:
Why would I do A versus B? You see the same thing on investments. Do I want a Roth, an IRA, a 401(k), a Roth? There's all these things, an index fund, that you see on Robinhood. And they'll be like, "What should I be doing, Jeff? What's a savings account, a CD, and a mutual fund? What is the right choice? How do I think about what I want to do?"
So I feel like that at the beginning of the flow when somebody comes to you and goes, "I got a house, I found it, I need a mortgage." They passed this part. They've got it, they want the 15-year fixed, whatever. But I think often, many consumers come to us and they're not quite sure what they're doing. They've got a problem, they don't know the right solution.
They've got a question, they want help navigating the universe of options. Then I find that once they've determined what they need, they want you to get out of the way. Most people, not always, there's also just difference between people. But typically once you're in the process of completing an application for a particular product, whatever kind of product it might be, people want to self-serve. They want to sit on their couch and they don't want to talk to a human.
Maybe cashing checks is the ultimate example. Nobody wanted to go to the branch and do it. Nobody even wanted to go to the ATM to do it. They just wanted to sit on their couch and take a photo. And it's better for me in the middle when I'm transacting. And then I think the end is the beginning, which is A, if I have challenges, let's say with making loan payments, I'm back to that position of saying, "Hey, I might need some help or some clarity," or it's my next product, which is like, "Hey, this saving's account's going great, it's gotten pretty big. Should I be doing something else?"
Which is the beginning and the end. It's the end of one product or the end of the life cycle where you've gotten into a product, you're using it successfully, and now it might be the beginning of another cycle. So being able to be accessed in that moment when things are going well, hey, I've got this one loan, I'm paying it off. Well, now I'm thinking about buying a car. That's the beginning of another thing, but you see it because you're interacting with them through the context of maybe the mortgage they have or the bank account they have.
So I think it's that, the beginning of the end. And then I think it can vary by person. We still see some people who want to talk to a human being. They don't trust the tech and they want to feel like they talk to a human who's answering their questions. And I think the more you can get out of the way of letting people who want to self-serve self-serve, the more you can put your resources towards those people who either were at a moment in time when they need the human touch or who just for personal personality reasons or whatever feel like they want to have that touch at some certain point in the process.
James Robert Lay:
I think that's where let's dive a little bit deeper into the opportunities around technology. And it could be through two lenses. It can be through self-serve, or it could also be from more of a proactive stance utilizing data AI to identify opportunities to optimize the lending experience.
Jeff Keltner:
Absolutely.
James Robert Lay:
And I want to cover both points, AI and data. But I want to start with data, because I look at data as the oil of the digital growth engine. Oil helps to make the engine run smoothly. And where do you see opportunities, this idea of running smoothly, reducing frictions? Where do you see opportunities for financial brands to use data to remove friction from the lending process, the lending experience, both online but as well as in branch? Because I think we need to keep both channels in mind right now.
Jeff Keltner:
Oh, so many places. I actually don't like the analogy of data is the new oil. I'll tell you why. I feel like oil is one of those things that most people don't have and the presence of it is valuable to you. And data is the polar opposite in many ways. Everybody's generating data all the time, and having access to data isn't the problem. It's the refining part. It's the figuring out what my data tells me and how to use it well that's the really hard and valuable part.
I think every financial institution generates so much data and yet they don't actually know how to mine it and refine it and turn it into useful insight. So I think there's a lot of value in it, but I also think a lot of times, it's not finding the data that's useful. It's like, how do I leverage it?
James Robert Lay:
Well, I think that's a great point. I want to address that real fast, because I think data is ones and zeros. Analytics begins to visualize that data, dashboards, et cetera. But then the most important practical point of this is to turn the insight into something that we can take action-
Jeff Keltner:
That's right.
James Robert Lay:
... or provide recommendations around. So I like that point that you're making right here.
Jeff Keltner:
Yeah, I totally agree. So if you ask where can we use it, I mean, the number one most easy thing is what is the product that this customer is most likely to be in need of right now? I can't talk to all my customers all the time, and if I could, I certainly couldn't talk to them about all my products.
But what is it that Jeff needs or James needs? Today, what should I be talking to them about? Maybe it's nothing. Maybe it's a savings account, maybe it's a car loan, maybe it's an investment, I don't know. But if you can reach the right person through the right medium with the right message at the right time, you've got a lot of opportunity. That intelligence could be what I'm marketing to them on Facebook. It could be what I'm putting in an email campaign.
It could be what I'm putting in front of a customer service agent in a branch to talk to you about when you come in for something else, like opportunities, things that they might think about. "Hey, I'm looking at your car loan and you're paying way more than we would offer you. You should refinance that." "That's crazy."
"Hey, I'm looking at your credit cards and I can see on your credit file, you're carrying $20,000 in credit card bills. Have you thought about refinancing that with a personal loan?" Whatever that might be. When I do these things, I start at the top of the funnel. How do people find product? So I start there to use data. How do you get the right product in front of the right person at the right time? You go straight down to the area Upstart really focused, which is the belief that one of the frictions of the credit system is that we say no to a lot of people as an industry.
And data will tell you, since we're talking about data, we did a little research with some of the credit bureaus and found that 80% of American consumers who've taken out a personal credit obligation have never defaulted on it. And yet less than half of American consumers have a credit score that would qualify them for prime credit. So what that means is when you use credit scores or I would argue most traditional approaches to credit, you may achieve the loss rates that you want, but you do it by not approving a large number of people who would have paid back given the chance.
And that's friction in the process. That's lost economics for an institution that's poorly served customers with a bad experience and a suboptimal relationship. So I think second place you use data is understanding better who you can and should approve for loans of different sizes, and how you should price those. And then next in line to me is how do I get you through the process?
Things like how do I do KYC and ID verification? How do I get comfortable with income? There are so many data points available other than what we've traditionally done. So often, I see institutions digitizing a legacy process. I go, "Oh, we've got a digital loan now."
James Robert Lay:
That.
Jeff Keltner:
I go, "Okay, how do you do ID verification?" "Well, in the branch, we looked at your ID. So now you take a picture of your ID." "How do you do income verification?" "Well, in the branch, we looked at your pay stubs, so now we have you send us your pay stub." And I go, "Well, have you thought about doing something that could have lower friction for the borrower?"
And there are a lot of data to do that. Even we talk about the data as the new oil concept, or is it, often institutions don't even have an automated way to look at direct deposits into their own depository accounts to validate a stated income. That's the easiest kind. That's your data. There's ways to get it from third party institutions, but even if it's your customer, I was asked when I applied for mortgage, "Please provide three years of tax returns to validate your income." And I said, "You've been my bank for 20 years. I've direct deposited every paycheck into my deposit account. You shouldn't ask me that." And yet they did.
So can you take that friction out of the process? And then I think that the last one is when you're in for a loan product, when you're in servicing, where is someone maybe in a financial distress? Are there signals about what communication mechanisms work best with different consumers who might be in need of a hardship? Maybe I want to change my outreach, proactively reach out to people who might be struggling about forbearance programs if we're in difficult economic times.
So there's all sorts of ways you can use data to optimize all the way through there. All the way back to as we said, the end is the beginning, that customer who got one product, what is the next product that I should be positioning in front of them? Back to the beginning of the cycle.
James Robert Lay:
Well, that's where I think combining first party digital data, i.e. website data, back to customer data that resides in an LOS or a core, and connecting-
Jeff Keltner:
Or CRM.
James Robert Lay:
Or CRM. Connecting all of these dots together, because you can pick up buying signals from someone's digital exhaust on a website.
Jeff Keltner:
Totally.
James Robert Lay:
And make some predictive measures, and I think it's transforming the operational model in banking, which has historically been reactive, waiting for someone to walk into a branch, waiting for someone to take some type of action, applying online. And to your point, it's like taking archaic processes that were built for the physical world and then just re-engineering them for the digital world. It's not an optimization. The mechanism is the same. I forgot where-
Jeff Keltner:
That's right.
James Robert Lay:
... I read this. But it was a data point, that for every 10 seconds added to a digital application experience, it decreases by X percent of conversion. So what are we asking that we don't need to be asking? And I'm curious, it's like, okay, so we have all of this data at our disposal. We have the ability to use AI and ML to get an augmentation and a capability upgrade. What's holding financial brands back from maximizing this potential?
Jeff Keltner:
Well, I'll say too things. I think having all the data and having it at your disposal are two very different things. So often it is can I actually connect the data from my website or my mobile app usage with my LOS, with my CRM, and actually make sense of that data to understand those signals in conjunction?
And the answer is often that architecturally, infrastructurally, and the way our technology systems are built, it might take three weeks to have four different analyst groups combine those data out of different databases and collate them with some ID record. And then it's like, "Oh, yeah, we can get you that data in three weeks."
And you go, "I want it in real time in front of my agent. I want it in my targeting algorithm for an ad campaign." So I do think I've often advised the institutions I talk to that one of the areas I see of under-investment generally is the plumbing so to speak, for data and technology. Because it's so easy to focus on did we launch a digital application, and you don't ask the question of were we able to connect it to... I go back to my example of looking at your bank transactions to validate your income or target an ad, or even as simple as when I've applied for my mortgage, my bank said, "What's your address?" I said, "The same one you sent my statements for the last 20 years. You should at least have some idea."
But those systems aren't integrated. That application system was a brand new system that was launched, wasn't integrated into a core or CRM. So figuring out how to weave these things together and give a unified view across them of the data that you have and then to start to glean insights on top of that data, having it and being able to use it in that way I think are very different. And investment in the plumbing of connecting these things together and making that data available to the right users at the right time is really important to be able to take advantage of these things.
It's not easy and it's often siloed by our business units. This system was bought by the guy who runs the mortgage business, which is different than the guy who runs the credit card business. They weren't designed to speak together because we ran as functionally independent groups in many ways. And now you're saying, "Well, we should be able to glean insight from A and apply it to B." We got to plug those two things together.
So I think that's one of the biggest things standing in the way. And then I think the other, and this is maybe not something our audience can do something about immediately, but is in certain areas take underwriting, there is some lack of clarity around how certain regulations apply in the context of machine learning model or the usage of alternative data points.
And I think they'll be increasingly asked for clarity on what's in and outside, certain policies or the way we approach let's say fair lending testing was built on the assumption of a scorecard model. And when you're using the AI model, the scorecard is not there. The same question applies, but you need to answer it in a different way. I think sometimes our institutions are conservative in how they are willing to push the boundaries on the new.
So they tend to say, "Well, let's wait and see how this plays out over time for other folks." So there's a hesitance to be first, but I think there's a lot of benefit to being early in these things, because you really differentiate.
James Robert Lay:
Well, that's a great point you make on the compliance side, because I know when it comes to say, AI and lending, there's a big concern around bias in AI models considering the fact that many financial brands will only lend to those that have a credit score of 700 plus. So what are the opportunities here?
Back to your point about not just waiting, but testing, applying, learning, looking for financial brands to increase the inclusivity of lending by eliminating bias through AI and ML?
Jeff Keltner:
I love the question, because I think sometimes we get so wrapped up around fairness and fair lending. And I'm not saying that fairness is a bad thing, so don't let me come off the wrong way there. But I think that fairness can leave out the concept of inclusivity and the idea that if we rely on traditional metrics, the typical African American does not have as a high credit score as a typical Caucasian American. Same thing is true for Hispanic American. So if we're just relying on credit score and saying that's already a system that is by and large leaving out disproportionately minority communities from access to credit.
And I think that AI and a better understanding of credit can help us bring those communities back in. We of course want to do it in a fair way, but I think sometimes the way we think about fairness can... Say you give the same approval rate for this population and that population. Well, if you're using credit scores, you don't know, you might have a different approval rate on your applicant pool because you just only marketed to the people with a 700 plus.
So it looks fair, but it's not really fair because a lot of people have never even bothered to take the first step and get into the denominator of your test. So are we really measuring fairness there? But are we really measuring inclusivity, which is where I think you should be. There's a huge opportunity for AI, alternative data, just new approaches to underwriting to identify those people who traditionally wouldn't get access to credit, be given access to credit, and show that they are credit worthy and deserve that access to credit and close that inclusivity gap.
So we believe there's a huge opportunity, and I will just say, one of the first things Upstart did before we even started lending was we called the CFPB and said, "We think we can close this inclusivity gap, but there's questions about how to model fairness." And to their credit, the regulators were quite willing to engage in a discussion about how to think about the concepts of fairness in the context of a model like AI, and we built some testing in conjunction with the Bureau, and I think that kind of effort is there. And the regulators frankly, they care about inclusivity.
I mean, they asked us to publish metrics and send them data on how much we could increase approvals and decrease cost of borrowing, as well as of course how that was being applied across demographics. But I think there's a huge opportunity for these things to open up the inclusivity of the system to people who traditionally haven't had the same kind of access.
James Robert Lay:
Well, that's where looking at some of this data, the Upstart model versus the traditional bank model, there's 75% fewer defaults at the same approval rate when it comes to the Upstart model, and then 173% more approvals at the same loss rate comparative to traditional incumbents. You mentioned something before about alternative data. What should the dear listener be considering here around alternative data when it comes to increasing inclusivity through lending?
Jeff Keltner:
When I think about alternative data, sometimes it's a scary phrase for people. I was just on a webinar with Equifax and some other lenders about this topic, and I think it almost sounds like you're scouring Twitter. Maybe not Twitter these days after what Elon Musk is doing with Twitter, Facebook, Meta, or whatever, and looking at these weird signals.
And I just think there's so much data that's not traditionally used in underwriting, but it's so clearly financial in nature. The kind of job you have, the kind of income you have, transactions in your bank that might be able to indicate income, just cashflow based underwriting, even if you just think about the credit file, most lenders get a handful of variables, maybe a dozen off the credit report. We get over 1,000. And it turns out when you look at all the 1,000 in detail, you get a much better sense of credit worthiness.
And then there's things that very clearly ought to be in some way in a person's credit history, which is have they been consistent in paying their utilities and their rent and their cell phone bill?
James Robert Lay:
Yes.
Jeff Keltner:
And some of those, we're starting to see come into the credit files. But those would all be alternative data points today, and yet I think are clearly financial in nature, clearly have something to tell you about the credit worthiness of a consumer.
And I usually think of this as can you find other ways to demonstrate credit worthiness outside of the credit score, which is typically based on a history of repayment. I don't really care about your history of repayment. I do, but I really care about what it says about your future of repayment.
James Robert Lay:
Future.
Jeff Keltner:
And there are other ways than a history of repaying loans to demonstrate your likelihood to repay. And we're looking for all those signals. Where are there some signals that can give us a positive indication of someone's likelihood to repay that we can use to extend credit?
James Robert Lay:
That's a fantastic point, because if we're talking about there's been a big conversation around financial education, I think it's only one half of the equation. I think financial empowerment or increasing the financial confidence of people through increasing their financial competence, that's more of a holistic view.
Back to the point of the internal silos, it's almost like you've got a cardiologist that isn't connected to your neurologist. I want both parties talking here, because it's the holistic picture. But then there's this future focused piece of this as well, because a person's past I don't think should predict their future. It should be the actions and behaviors that they're having in the present moment as a trend towards a bigger future for them.
And I like the perspective of alternative. That's why I wanted you to clarify this, because I think when you hear the word alternative data, you go, "Well, what is that?" But I mean, cell phone bills, utilities, rent, those are all predictive patterns of people who might not necessarily have a positive "credit score" that says, "Hey, I'm making progress towards creating a bigger future. Can you help me out? Can you get me there?"
Jeff Keltner:
Yeah, that's right. And there's so many ways that could be happening. To your point, obviously all of these are past signals. You paid your rent, that's a past action. It doesn't necessarily mean you'll pay next month's rent. It's probably a pretty good sign. It's interesting because they're alternative data points, yes, but in some ways, no, which is to say many credit policies, when we have a human loan officer, they often make exceptions and they have these phrases in the credit policy.
Unless there are other substantial compensating factors, compensating factors, and one of the ways I've thought about, and this isn't strictly technically speaking the way that a system works, but one of the ways you can think about alternative data is a systematic approach to compensating factors as opposed to saying, "Did the loan officer ask these questions? In their judgment, were there compensating factors?" If your credit score's a little below the bar, this one was enough. But if it was farther below the bar, maybe you needed to have three things. There's not a systemic approach to using compensating factors, in a traditional way.
Traditionally, it's a human judgment. And you could in many ways think of alternative data and machine learning as a statistically valid way of looking at a universe of compensating factors, figuring out which ones actually tell you enough about likelihood to repay to overcome... And then you really say, "We're just changing our credit policy," as opposed to saying, "We have a hard cut at X," we're saying, "No, we don't have a hard cut at X. We're willing to consider other factors that allow us to go below a certain credit score or a certain debt to income ratio, when these things are true."
And we're not using our intuition about which ones should be right or our personal relationship, we're using statistics to tell us which of these actually are substantial enough in predictive value that they can outweigh a low credit score, let's say, in telling us that this person is still likely to repay. It's a statistically valid way of saying, "Yeah, these things actually outweigh that thing, and we should therefore be extending credit to this person."
James Robert Lay:
It's a great point about the idea of "relationship lending," which I know a lot of community banks, credit unions even have used that term over the years. This is statistical. We're really relying on the data here through AI and through ML to provide a lot more clarity going forward into the future.
Speaking about the future, I want to stay future focused for just a bit, the world of lending, it has transformed, it will continue to transform, especially when it comes to these two words: embedded finance. The big trends, and I think as we're looking at future focused, what are the big trends that financial brands must be aware of and really paying attention to, really watching when it comes to optimizing the purchasing experience for consumers?
Jeff Keltner:
Well, I think the key words of those last ones, the purchasing experience, I mean, I think increasingly, embedded finances I think is one of those... hides a lot of sins. People use it to mean a lot of things and it's not always exactly what we mean when we say embedded finance, but I do think that the idea that the technology now allows us to embed a financial experience closer to the point of transaction, be that things like earned paycheck advance for things where I'm moving my financial [inaudible 00:28:25] closer to my employer. I'm moving my paycheck closer out of my bank and into my employer experience sometimes.
Or whether it's buy now, pay later, or embedded finance on the purchase side, we've known this is true for a while. I mean, it's interesting, but no one when they want to buy a house goes to the bank first. They call a real estate agent, that's the first call they make.
And when you want to buy a car, you also don't call your bank and go, "Hey, what can I afford?" Maybe you should, but typically we would go onto the website of the OEM and check out what we want to buy or we go to a dealer and see what's available to purchase, and I think we increasingly live in a world where you can push the financial transaction closer to that point of purchase. The ultimate, almost all financing it should be said, outside of refinancing, personal loans are often refinancing. But most things, I don't get a car loan because I want a car loan, I get a car loan because I want a car.
I don't get a mortgage because I want a mortgage, I don't want a mortgage. I want a house. And the mortgage gets dragged along, and the more we can move the totality or as much as possible of the length of lending and borrowing experience into that transaction of the thing I'm actually trying to do, the more likely that is. And we think of it as a new thing, but if you bought a car at a dealership in the last 30 years, you've experienced the original form of embedded financing because the lending was taking place at the car dealership, you walked out of there with a loan from a local bank.
So I think that that becoming reality digitally makes a lot of sense. But also in the context of physical retail locations or I mean, it's happening in healthcare. Dentist offices, other places, and I think the idea that you won't necessarily come to a financial institution to transact financially. You might do it in the context of another activity that you're doing, be it purchasing something or you can imagine this on the investment sides as well, where that's going to become the reality.
And to me, that means a bank needs to have... saying an API enabled approach is bad, but needs to be thinking about how they can take their experiences outside of their own owned properties, your website, your mobile app, and embed them either through APIs or embeddable widgets or some sort of experience into different mediums of transaction. I think that's just the reality of what's going to happen. And the people who can do that effectively are going to win the business that happens there.
James Robert Lay:
And that creates an exponential growth opportunity, especially for community financial brands that historically have been confined to growth based upon boundaries, zip codes, and borders, and cities. But now when you look at embedded finance, and I think of one organization in particular who has been in our program for the past couple of years, they're a community organization, but they're making now a national play through dealer direct in a couple of different niche verticals and they're building up the expertise within those niche verticals.
So it's becoming a B2B2C model, working with these dealers within these niche verticals to provide almost that real time financing at the point of purchase based upon the products that these dealers are selling to solve the common pain points of people that historically you wouldn't think about going directly to a bank or a credit union to get financing through.
It's reducing the overall friction. So I like this idea, and I think the key point and the takeaway of all of this is continuously be a learner here, because the world is evolving. Knowledge is key, staying aware. And that's one of the reasons that I want to bring up your podcast, the Leaders in Lending Podcast. You've got an audience here that might also find value in what you're sharing, and if you think about some of the recent conversations that you have had on your podcast, what's maybe a big insight or a-ha that you can transfer from the Leaders in Lending Podcast to the Banking on Digital Growth Podcast audience?
Jeff Keltner:
Well, I mean, I think it's interesting. So the Leaders in Lending Podcast, I typically talk to bank and credit union executives. Sometimes now, I'm talking to more and more FinTech players, and I think this conversation about embedded finance comes up a lot. But I think the insight is one of our guests said it best, I think he said, "If you were thinking about a digital transformation without first thinking about a process to re-engineering, then you're getting it wrong." Don't digitize the legacy process.
And to me, that's the key thing that the real leaders are seeing and doing. And embedded finance to me is part and parcel of that, in the sense that I think as you engineer your current solutions, I often talk to our teams about when you're architecting a solution, there's choices that really impinge on your future optionality and there's choices that don't.
And I don't care when we're engineering if we have code that gets thrown away in six weeks or six months, that's okay. Six weeks may be a little tight, but six months is okay. But if we are architecting so that it's really hard to do one of the things we think we might need to do in the future, that's really tough. Sometimes you make a choice that says, "Hey, this thing that could've been a six-week project is now a six-month project, because we did data engineering wrong or we architected our system wrong, and we can't support that use case."
And I think embedded finance is one of those, we don't know exactly what the winning use cases will be. And that's the thing, you got to architect the idea that there are use cases you haven't thought of into that. So that architecture level, the plumbing level thinking about how to build out your systems to support different modes of transaction is really important.
And the idea that digital is not the goal. The goal has got to be improving the customer experience and digital is a part of that, but not the only part. There's a process to re-engineering. I talked to guys, HELOCs take 90 days and it's not because it's not digital. If you do a digital HELOC, that still takes 90 days, but it's all on a screen. You haven't really solved the pain point the customer had. So if you're taking that customer-centric approach, then you start to go, "Well, how do I change the 90 into 10, and maybe use technology to do that? But my objective isn't to make it digital. My objective is to improve the experience of my customer." I think that's the key thing.
And I get lots of examples on the podcast of how people are thinking that way or doing that. But that's I think the core insight I've taken away from the conversation.
James Robert Lay:
Well, it comes down to putting people at the center of all of your thinking and doing. It's process. I like that, it's process transformation, solving the common pain points causing common people problems. Let's wrap up on this. There's a lot of opportunity. Where do we get started? What's one small step the dear listener can do next to... Maybe it is process transformation, to just optimize the digital living experience to maximize future growth?
Jeff Keltner:
Well, here's my one tip to get started, and it's surprising, have you done it yourself? Do you know what it actually is like for a customer? I had a guest recently and she came on, she was new at the institution. And she said, "I just went on and I tried to open all the accounts," and it's so rare that we get caught up in our part of the business that I'm optimizing the call center or the digital experience or the marketing funnel.
And how often do we really have a good understanding across our teams of what happens? So much of what causes customers pain is at the edge of these things. It's where the engineering team makes the credit policy team. And if you're sitting in the engineering team or you're sitting in the marketing team or you're sitting in the credit policy team, you don't see it until you take the customer's point of view.
So I would just encourage everyone, we actually started doing this for all of our product teams, they have to have an internal video demo. Because some of the things like a credit product, you can't actually go to the end, because then I got to get a loan. And I've done that before to test the product, but everybody may not want to do that.
But the idea that we can take any one of our products and go through a video experience of what the consumer experience is from A to B and what that looks like and see it and understand it, to me that's the beginning of everything. On Upstart, I used to go through that process three or four times a week, just constantly going through it. And say, "Hey, we changed it." I go, "Okay, now I've frozen my credit now because of all the hacks and it's harder to do because I don't want my credit scores to come up."
But the idea that you are centered on your customer experience and to be centered on it, you have to know it. You have to see it and be able to actually tell people, "Here's how it works." That's my first step, because as soon as you do that, you'll go, "Here's five things that seem broken." And then that's your next step. It's like, what are the things that seem broken and how do we make those better?
James Robert Lay:
Well, I think that right there, it's so practical and it's one of the reasons that we have now conducted over 1,200 digital secret shopping studies for financial brands on the front end experience. So i.e. website, product positioning, application process up to the point of hitting the actual submit button to your point of getting all those hits on credit.
But there's a tremendous amount of insight, I mean, even looking on upstart.com and the way that this is positioned, get a smarter loan, checking your rate won't affect your credit score, what would you like to do? And it's almost like the experience of someone walking into a branch. "How can I help you?" "Pay off my credit cards, consolidate my debt, refinance my car, something else," and then one clicks on one of those CTAs, and then it asks a question, "How much would you like to borrow?"
And it's one question per screen, which we have found that by asking one question per screen increases the likelihood of conversion because you're reducing the cognitive load of what is being asked. So therefore it doesn't feel so overwhelming. So I think this idea of continuously learning, testing, refining, optimizing is a great way to look at process engineering or process re-engineering.
Jeff Keltner:
I will say you're totally right. We used to have one big form, and we now have... I think it's on mobile and desktop, one question per page. And there's so many instances like that where you'll have a debate about will A convert better or B convert better? Will customers like this or that? And I think I'm pretty good at this stuff.
And what I know is that I'm not almost always wrong, that would be improbable because it would be as hard to do as being always right, but that your users will surprise you. So things like that may be counterintuitive, but there was a saying that's attributed to Marissa Mayer, Google said, "In data we trust." So the idea that, hey, we don't settle disputes here based on like, "Well, I theoretically think..." and the consumer and you can make the argument of cognitive load, and ultimately the argument of is it too much cognitive load or not is not won by an argument. It's won by data.
It says, "Well, let's try them both and let's see which one converts better, and that will give us real information." Then you can understand why and maybe apply that logic to something else to make a guess. But the ability to test and iterate is really critical because the only thing that I know for sure to be true is that I will be surprised by the results of tests in the future where I go, "Huh, that's not what I would've thought." But if that's what the data shows me, then I'm going to follow the data down the path of what makes the most sense for my customers.
James Robert Lay:
And I think that's why you, when starting every type of optimization test like this, enter in with a bit of Socratic wisdom. I know I know nothing, and go in with an open mind. In data we do trust. So good thinking, Jeff. Thank you so much for the conversation today. What is the best way for someone to reach out, say hello, continue the discussion we started here?
Jeff Keltner:
I'm on several social media platforms, but mostly LinkedIn, so you can find the company, upstart.com. You can find Leaders in Lending wherever you get this podcast and your other favorite podcasts, and I'm pretty much easily findable and communicable on LinkedIn, so I look forward to connecting with people there.
James Robert Lay:
Connect with Jeff, subscribe to the podcast, listen to Jeff, learn with Jeff. Jeff, thanks so much for joining me for another episode of Banking on Digital Growth.
Jeff Keltner:
Thank you for having me.
James Robert Lay:
As always and until next time, be well, do good, and make your bed
