Fair Isaac Corporation ($FICO)

All right. Good afternoon, everybody. Thank you for being here. We're happy to continue our session here with FICO. We have CEO, Will Lansing; and CFO, Steve Weber. So thank you both for being here.

Well, maybe -- I guess, let's take a step back. I think there are some newer investors in the space now. Obviously, your stock has been under some pressure for a while. So maybe kind of from your perspective, what is the FICO pitch today from your perspective? Because you guys keep obviously buying back shares every opportunity you get.

Yes. Well, I guess just on that point, we're kind of a slow-moving LBO. We have been buying back our stock for a very long time. We started at 74 million shares. And when I joined the Board, we were at 36 million shares. When I became CEO in 2012, we were at 30 million shares. Today, we're at 23 million shares. So what we do is we take our cash flow and we buy back stock and lever up and our cash flow grows, and then we buy back some more stock and lever up a little bit more and trying to keep the leverage up in kind of a 2 to 3x range. So that's the capital side of the equation. And then the business has just gotten stronger and stronger and stronger for -- well, certainly for the last 14 years. And we see a pretty bright future. We're a little surprised at where the stock price is, frankly. We're kind of trading at multiples that surprise us. And maybe it's just worth going back in time and giving you a little bit of the FICO story so you kind of have a level set on where we are. But I guess before I do that, I'd remind you, take a look at the charts, we had well over a decade of kind of rocketing to top decile performance on TSR. And then in the last 15 months, we've been pretty beat up mostly because of fear about some political things going on in Washington and our mortgage business. So I'm sure we'll spend more time on that. But again, by way of context, just maybe a few minutes on FICO history and how we got here. Well, actually, just a show of hands, how many of you guys know the story, the FICO story. You guys -- you guys know nothing about it. So -- that's fine. So it's a good place to start. The company was founded in 1956 by a mathematician and an engineer. And the idea was let's apply analytics to data and make better decisions. And so for 30 or 40 years, that was the business. It was a consulting business, a body business of doing that. We built scorecards, proprietary scorecards for banks, because banks had big money decisions, it was worth taking the time to invest in a better decision. Now went on. In the 1970s, we got this idea that maybe we should try to build a software business too because if we have software, if we embody the analytics and software, we'll be able to get returns to scale, make money while we sleep, we won't just be a body business. So we started a software business. We took the half dozen questions that banks asked us most frequently. And we built applications around them. So originations, should we or should we not make this loan? Customer management, line management, should we increase this person's credit card line? Should we decrease it? If we increase it, should we decrease something else, some other risks somewhere else? Fraud, fraud detection, collections and recovery. So these were the big areas where we had applications and we've built a really substantial software business in the -- through the '80s, '90s, 2000 and on around these franchises. I took over in 2012 and we made some changes. On the software side, we went to a cloud model from an on-premise model. And then we also put a lot of energy into building out the technology stack so that we're not just an applications business anymore. We are a decisioning platform. We're a true platform. And it took a lot of years to achieve that success, but that's kind of where we are now. We'll spend more time on it if we have time at the end. On the Score side, in 1987, we decided instead of building proprietary scorecards for each bank, what we would do is build a generic scorecard that could be used by any bank. They just show up. They don't have to give us any special parameters. We did that in partnership with Equifax and it was wildly successful. We kind of made this thing that made it super easy for lenders to evaluate credit, low-cost evaluation of credit. And this is also in response to the fair lending laws that want you to demonstrate that there's no discrimination in the way you do your underwriting and scores are completely clean and science-based. Then we had this idea that we should build the same kind of a score with the other bureaus. So there -- in the U.S., there are 3 bureaus. Equifax experience and TransUnion the big 3. And so we went and built scores with Experian and with TransUnion as well as with Equifax. And what we did was we aligned the ODS score ratio across all 3 bureaus, the same. So what that did was effectively say, FICO scores a FICO score, regardless of where you get your data, regardless of which bureau you use. You can imagine how popular that was with the lenders. I mean, to commoditize the data at some level. and it gave them a lot of leverage versus the data providers because they could easily switch from one bureau to another. So pretty popular. That went on and we have the lenders liking it. Regulators got into it. They thought that it was pretty useful for them because they can measure the risk of the banks that they're regulating what's the average FICO score of the portfolio, how we will behave in a downturn. And then the securitization market evolved and they needed a metric for pricing the risk and the most obvious one was FICO. And so pretty much all your asset-backed securities, your mortgage-backed securities all come with an average weighted FICO score so that people actually know what the paper is worth. And so now you've got lenders, you've got regulators, you got the securities, more of the investors in the securities market. And then finally, we got consumers on board. We give them their score for free. And so there's a lot of demand pull for FICO scores. So that's the -- that's how Scores got to be the industry standard that they are and kind of positioned us where we are. Another wrinkle starting in about a decade ago, we started raising prices for scores. Why did we do that? Well, for 30 years, we didn't raise prices for Scores because our pricing was hardwired into our contracts with the bureaus. We distribute our scores through the bureaus, and so the prices were all fixed, and we never changed them. And you should just talk to my predecessors about why? And I decided we should have the flexibility to change our pricing and 30 years is too long to go without a price increase. And so we started changing our prices about a decade ago. And so every year, we're in the business of trying to capture some of the value that's been left behind. I mean, that's really the story of FICO is -- we have a mountain of IP that has been undermanaged and undermonetized for decades. And we're in the process of figuring out how do you monetize that with the least amount of disruption to the markets in a steady and continuous way, march up the value -- march up the monetization of the value. And so that's what we've been doing. In mortgage, in particular, we've raised prices, if you read the bad press that they focus on the percentage increases that we've raised prices, and it's true, we've raised prices from $0.05 years ago to $10 today. But remember, that's $10 out of $6,000 of closing costs. So it's still kind of a trivial number in the scheme of things. But what's happened is it turned into a headline, our mortgage price increases turned into a headline for the populists in Washington and in particular, for the Director of the FHFA, who's very pro competition, very focused on how do I make housing more affordable in the U.S., one of the things he's looking at is closing costs. And so among other things, he's questioned our pricing. And I think that we've actually responded by coming out with some new pricing models, some innovation that I think he likes where we have an option to buy mortgage squares from FICO for $0.99 -- so that's a big step in the direction that I think he was looking for. But I would say the noise in our stock is almost entirely tied to that. There might be a little bit of spillover from the AI software scare -- but I think that's pretty minor. We never got much credit for software before. So I don't know how much we should get back for it now. We are very happy with the software business grew 49% last quarter in the -- in our new platform. So strong business. So that's the background.

All right. Helpful. We'll get into some of the noise, but maybe, Steve, you can help here. But like with that history, with the current context around pricing and mortgage and so forth. Is there a long-term algorithm, financial algorithm that the audience should keep in mind whether revenue EPS?

Yes. I mean we don't have any issued numbers. But if you look at our past, you can kind of see that we've been pretty consistent. So we do target internally. We want to maintain the same pace that we've done in the past. So we try to get in at least into the teens of revenue growth, which could drive us into the 20-plus percent with margin expansion in -- of net income. And then a lot of times, that can be above 30% with EPS with the buyback. So just rough justice. That's how we've been looking at it. And that's -- as we make plans for the coming year, that's what we try to target going forward.

Got it. Okay. Well, let's jump into the mortgage market and the political noise you're referring to -- so we just had MSCI before you. We had S&P in the morning. those indices, there's a lot of competition, but bunch of the benchmark and the benchmark. What we described it FICO is the benchmark. Obviously, direct ability has implemented a lender's choice allowed or in the process of allowing VantageScore to come in -- from your perspective, how do you see the potential shifts in the market if and when they are approved.

Well, we'll see if Vantage gets any share at all. I mean, we've been competing with Vanish for over 20 years in every other market outside mortgage and they have no share. They push a lot of scores and the way they do that is by sending them along for free. So if a lender or as a bureau for a credit file and a FICO score, they'll often get the Vantage score sent along for free. And that's where the big numbers come from. But I'm not exactly sure that should be considered market share because nobody asked for it and nobody paid for it. in mortgage remains to be seen whether having an option for Vantage score get you anywhere. I would say that the value proposition is still pretty thin. So why would someone buy a Vantage score if they're buying it because they're interested in credit default risk and predictiveness and certainly, the entire nonconforming market cares about credit default risk. And I would say that the conforming market also cares about credit default risk. -- because it comes back to haunt them if the loans go bad, it -- a lot of those loans are put back by the GSEs to the originator. So they also care about credit default risk. If that's what you care about, FICO 10T is 8% more predictive than Vantage 4, and FICO 10T qualifies 5% more borrowers than vanish. So predictiveness is not going to get you there from a value proposition standpoint. So then you say, well, what about on price? Maybe somebody cares about price. And even though we're talking about a very small price in the scheme of things, is there someone who cares -- and on that, I'd say our new pricing model at $0.99 meets Vantage score pricing at $0.99. So I don't think there's a reason to move for price. So that leaves you with one and only one reason to buy a Vantage score, and that's if you're trying to game the GSEs. And that's real. -- that's not hypothetical. I mean, any time you have a 2-core system, 1 score is always better than the other. So when you -- typically, you would give a better price to the better score. So if Fannie and Freddie are constantly being presented with the better of 2 scores, they're being gamed. And frankly, what will occur in the long run, if anybody does this gaming what occurs in the long run is that the nonconforming market, the lenders in the nonconforming market will skim off the best credits out of the conforming market and leave the with the inferior credits, which will cost Fannie and Freddie quite a lot. That is the likely outcome of gaming. But I would caution you just by saying that the addressable market for gaming is fairly limited. It's under 10%. So if you look at the rules that came out last week, which are you need less than 80% loan to value, meaning you have to put down 20% or you can't get the mortgage. So for Vantage, you need less than 80% loan-to-value. And you need -- there's not actually a separate Vantage LLPA grid. That's the pricing grid. There's not a separate one for Vantage. What they did was they said, take advantage for us to track 20 points and jam it into a FICO LLPA grid, which is suspect science, but anyway, that's what they've done. And if you work through all the math on it and I'll spare you guys the math today, it takes you to about a 9% addressable market.

And I guess, just to be clear, that even the gaming, in order to game, don't you just have to pull boat scores all the time so...

Well, you have a great share of loss -- great point. So there's .

There's no volume loss. So if you have to pull both scores in order to compare them to do the gaming, our volume doesn't go down. You're still pulling a FICO score. -- but the overall market of Scores volume went up a little bit. So yes, we could have -- technically, we get a share loss, but no volume loss. That's theoretically possible.

Got it. And then the $0.99, yes, you've matched kind of the upfront pull fee. But in the performance model, you do have the $65 closing fee per report. So how do you think the lenders, originators market will address that when they're trying to compare whether they want to save money or not?

Well, that's money that's paid by the consumer. Now ultimately, of course, consumers pay for everything. We know that -- so it's really about kind of what bucket you put it in. But the industry has always tried to push cost to the closing statement and out of their own P&Ls. And that's why we have respi laws. The rest well laws are designed to keep that from happening really to ensure that the only costs that go into a closing are appropriate costs that have something to do with the closing. . And to that extent, what we've done is match them perfectly. There's nothing that's closer to what the closing is all about, then you got your mortgage successful funding fee. So I think that we're actually helping the lenders with us -- in terms of their true cost to themselves, it's identical, whether they're paying $0.99 to Vantage or $0.99 to FICO.

Got it. And in order to get that $0.99 of FICO, they have to go through the direct -- the DLP model. .

That's correct. True you guys. That's right. It seems like .

It's been a case of going to be ready in the next month for a few months now. So what's the latest on when the DLP will be out there. .

It's going to be ready and Well, I mean, it shouldn't take that much longer. We're waiting on certification from the GSEs, they're supportive, the director is supportive. It's a matter of time. .

Okay. And all that you described on the lenders [indiscernible] takes care of the cost, once DLP is live, how fast and how much adoption should we be looking out for?

Well, no one really knows, but I think that there's a lot of demand for the performance model, which will only be available through the Tri-Merg resellers sellers through the DLP model. And our economic analysis says roughly half the market would prefer the performance model over the model. And the Tri-Merge resellers have by far, the lion's share of the market and almost all the scores are purchased ultimately through the tri-merge resellers. So it could easily be half it could get to half certainly within a year.

Got it. And I think it's probably worth spending a little bit of time on the 10 independent study that was put out recently because I think Vantesco,obviously, claims should be better than the FICO classic, which might make sense because it was developed so many years later. But can you just talk about this was a second iteration of the independent study, I believe. So -- what changed? And what are some of the stats again? And why 10s much better .

So I would encourage people to the FICO website, read our white paper on it and you can go to Millman, which is the independent third party that did the analysis and read theirs. It's quite technical. . But they've done a scientifically rigorous assessment of 10 T versus Vantage 4. And we are better on predictiveness -- they actually go through the gaming issues with a 2-core system and identify the costs around that. So I mean, it's worth a read, but it's quite technical. There's probably other things you'll read before you read that. Just take my word for it. FICO [indiscernible] Vantage for.

Okay. The FHFA had a press conference recently and actually 2 questions. So the first one is, I think the direct ability had mentioned that he had a conversation with you around this $0.99 pricing model. And then you had also mentioned a few times, it sounds too good to be true, so I'm going to make sure there's nothing in there. But I guess I just want to confirm, he knows that the performance model and the closing of the 65, that's not going to be new to him. .

No, no, no, no. He's a smart guy, and he's completely aware of the structure, and I think he supports it.

Okay. The other main thing for their conference is obviously the pilot. I guess they've selected 21 lenders to try and test out VantageScore. It's a little bit of a black box, great box, whatever for a lot of us. Anything your mortgage team and Julie may have shed any insights you can share with us.

There's not a lot to know there. I mean all those lenders, whoever they are, are under NDA, I can't talk about it and can't disclose. And so very little is known. We know some of the rules around accepting a VantageScore, the 20 points and the 80% loan-to-value. So -- but other than that, I can't tell you much I don't know. I think it's a manual process. I'm not sure.

Okay. Maybe the other thing you might be able to tell us about is the 10 data has not yet been released. They said in the coming months, summer. What is the holdback for when that gets released. Once that gets released, do you have to go to the 21 lenders if you go back and test that as well?

Don't know the answer to that. I know that the GSEs are working on it and intend to release. I mean, I think they mentioned in the press conference early summer. So we're kind of waiting for them to do that. they have their own processes and procedures, and they're working their way through them. We fully anticipate that it will be released, but we don't have a time line.

Got it. In terms of 10T, though, when it does come out, I think historically, you've talked about how even today, the mortgage market is using FICO version 4 and 5. So how quickly do you think 100 can be adopted or even VS 4?

Tent is being adopted fairly quickly in the nonconforming market. So it can be adopted. I do think there's a ton of inertia. I think if the -- if you ask the industry, do you want to change, they would say no, they would like to just stay with FICO classic forever. But we are seeing adoption in the nonconforming market. And then once all of Classic Vantage 4 and FICO 10T are accepted in the conforming market. It's a little hard to say. I mean you could see a shift. I mean there's -- if you were going to shift, you should shift from classic to 10T because you'd get more predictiveness out of it. But again, all the systems, models, everything is designed around classic. So we'll see how long that takes.

There's also bit of an advantage we can run it. I mean you run 10T in parallel with classic FICO, right? That's what's happened in the nonconforming market. So from that point of view, it's easier to run them both at the same time, compare the results and we do some forecasting on that.

Got it. And in the past, you've talked about it's a heavy uplift to switch from scores, but -- or go from the classic to something totally different like VantageScore. How -- what is the lift like to go from classic to 10T.

It's a little easier. So the reason FICO scores are designed to be backward compatible with prior generation FICO scores, and they always have been right up to FICO 10 and FICO 10 is backward compatible with FICO 9. FICO 10T uses trended data, so it's not perfectly backward compatible. It's a little bit different. But I would say that 10T is architecturally very similar to prior FICO scores. It leverages the same kind of weighting, same attributes -- and so the -- if you were going to make some assumptions about how 10T behaves relative to classic, 10T will be a lot closer than Vantage 4 would be. .

Got it. The $0.99 plus 65 for FICO 10T, I think, implies a modest price increase versus the 495 plus 33 you have for classic -- some people think that's your 2027 pricing that you've introduced today? Is that the case? Or how should we think about that.

No, no. I would say that 2027 pricing is not here yet. We haven't really decided what we're going to do there. Our our process is in September, August, September every year, we kind of review things, decide where it would be appropriate to make bigger pricing adjustments. -- and then we publish them to our partners so that they can be implemented on Jan 1. So none of that has occurred. Any pricing that you see in the market today is 2026 pricing?

Got it. And with all this noise with all this change, how should we think about what your price value gap in mortgages? Like let's say, it's roughly $10 a score today versus what you think it should be? Because I know you talked about this for years before this all began, I think people thought sure it's going to get there. But now I think they're a lot more skeptical. So is your approach change are you going to balance the annual increases differently? Or...

What's a FICO mortgage score worth, right? That's the question. Nobody knows the answer to that. I mean, really no 1 knows the answer to it. All you can do is look at analog. So if you look at S&P and Moody's, they charge approximately 8 basis points to rate of mortgage-backed security, a 2. So that's 16 basis points, okay? And that's typical. And those securities that they rate AAA rated, guaranteed by the government. They're all AAA rated. So I'm not sure how useful that is. But they also supply the average weighted FICO score of those securities. And that's what the pricing is built on. The pricing is built on the FICO score, not actually the S&P rating. And I only share that by way of background because we don't charge anything for using our score in that context. We charge 0. They charge 16 basis points and we charge 0 and yet the value is all coming from the FICO Score. Another way to think about it is 16 basis points on an average $400,000 mortgage, it's a little over $600. We charge $10 per square. So you guys will have to make up your own minds about what the value gap is, but I would submit that it's quite large and remain so.

Got it. Fair enough. Okay. Maybe one last opportunity on mortgage. Again, we've talked about a lot of noise. What do you think is the most underappreciated or misunderstood aspect of this debate that keeps pressuring your stock?

Look, I think that the FICO mortgage score is deeply, deeply embedded the system, not just because it's used by the originators to get the mortgage but everybody downstream uses it, that Fannie and Freddie used that score and their LLPA pricing grids, the mortgage insurers use it in their models for providing mortgage insurance. The credit risk transfer guy, CRT, they use it in their models. The mortgage-backed securities investors themselves use it when they're pricing the MBS the prudential regulators use it. So there are lots and lots of parties who are pretty deeply focused on FICO as the cornerstone of the system. So I think changing it's very hard, just really hard. I mean, you can create an option and say, sure, you have an option to use Banascore. Guess what? There's an option to use an score in every other market that we have in auto and credit card, in account management and free screen, you name it. You want a VantageScore and Vantage will be happy to provide you with one and yet nobody uses them. So how different is it going to be in the mortgage market, we'll have to see.

Got it. That's a good segue into the other parts of the market. So in auto for the last several years, I think you've done modest price increases and card has been inflation type stuff. So maybe just on auto, firstly, what -- like how do we think about the historical price increases? What's the opportunity going forward? .

Yes. I mean we've been, I think, like you said, modestly increasing prices in auto, A lot of it is really understanding the market, understanding the dynamics. There's a lot of different players in that space. There's a lot of value derived from the score. A lot of times, it's the score that determines whether a car is bought or sold. So there's a lot of value there to the dealers, the people that actually do the financing. So again, we just need to understand the market. And every year, we learn a little bit more and we look at the market and look for opportunities to raise prices there where we think that the value is much higher than what we charge.

Got it. And both in auto and maybe mortgage as well, right, does the forecasted volumes impact how much pricing you're going to take .

Yes, I think a little bit. I mean, because in the sense that we don't rely on increases in volume to provide our guidance. So every year, what we do is we sit down, we look at all the data sources come up with kind of a consensus view of the volumes for mortgage or auto or for whatever. And then we heavily haircut that because we don't believe that most of those are wrong. And because we're conservative. And if you followed us for any time, you know we have a reputation for sandbagging. So that's kind of where we get to. We don't count on a lot of volume increase to make our numbers. And then what happens is if we really do get a good volume year, which someday, we will have a mortgage for sure. we closed 5.8 million mortgages this past year. And the average over the last 5 years is over 8 million close mortgages. So we -- I mean, there's a lot of room to grow. Volumes will come back someday. But we don't count on that. Our guidance is not built on an expectation about volume increase.

We get questions a lot of times like if mortgage goes up next year, that means you do less in the price and said, well, we'll never know. -- right? So we can't count on that. So we tend to assume that the markets will remain relatively flat. .

Got it. And then maybe if we round it up like the card strategy, is that just inflation because it's a high volume market .

Yes, it's a high-volume market. It's probably there's probably more discretion. There's probably more elasticity because you really don't need the score in the same way, and there's not really the transaction that takes place that we are in the other markets. So it's more inflation. There are some pockets there where we will charge a little bit more than inflation, but it's closer to a CPI-type price increase.

Got it. Even when and if the DLP program is introduced the credit bureaus is still you're going to be your largest customer, whatever the disclosures you have in the 10-K -- what are the relationships like with the bureaus now with appeal .

Have improved a lot. I mean, look, it's not a secret that there was some friction. I mean we've gotten along very, very well for -- certainly for the last 14 years. And FICO surprised them with the direct license program. I mean, I called the CEOs of the bureaus several hours before we announced it to the world. And guess what, we're launching an alternate distribution channel, probably the most raddling thing in our industry in a decade. And needless to say, they didn't appreciate the lack of a heads up, the fact that we just dropped it on them. And of course, they had big revenue implications from them, big profit implications for them. So we caught them off guard. We didn't treat them with the respect that they deserve given our relationship. I mean so I understand why they reacted the way they did. As you know, they've all recovered, right? They figured out how to go get that revenue by increasing the data -- the price of the data. So there the hole that was anticipated has been kind of closed up. So I think that's resolved. And at the end of the day, we have symbiotic relationship with them. We need them as a channel. -- they need FICO scores or else their data is not worth very much. And so it's that combination that keeps us working closely together. And I'd say our relationships are pretty strong.

Got it. So no potential spillover effects into card and auto, any plans [indiscernible]. .

We don't have any plan .

Okay. And how about on the direct-to-consumer side, Experian obviously is a big partner there is that a completely separate entity relationship. .

They're a great partner there. No, it's not completely separate. It's part and parcel of the whole relationship. They are phenomenal consumer marketers. I would say that they have -- although I bring proud of my FICO in our own little consumer business. We give them every opportunity to take the business. We don't compete with them on search words. We -- I mean, we test things. And when they work, we tell them about it. And they've done just a phenomenal job building that consumer business. It's really impressive, and that will continue. We feel really good about it so today.

Got it. Okay. We have about 9 minutes left. Let's turn to software. Maybe a little bit of a not background, but a mix, like what is software. It is a little bit kind of many different things all over the place. So how would you simplify software amongst your top few products or mixes that will help us...

I think the easiest way to think about our software business is our new software business because the old 1 is the applications we talked about -- the new software business is a decisioning platform and it's kind of next-generation CRM. So the idea is take data from a lot of different places and apply some analytics -- and with that, make a decision, which you can then feed into a workflow, and that workflow happens in real time, it gets in real time that decision to a point of interaction with the consumer. So if you're a B2C company, whether you're a bank or a credit union or a retailer, anybody with a B2C relationship wants to optimize that interaction with the consumer to achieve some goal. And it's not always the same thing. It could be thereafter revenue, you're close to the ad quarter. And so they're all about conversion. Okay. So 1 shot up on my website. I just need them to convert and buy something and put a few more dollars in the revenue line or maybe revenue is fine, but profit is what matters. So let's emphasize for this customer, things that are more profitable for us, or maybe those are fine, and we want to focus on lifetime value. How do we get this customer from this category to a more valuable category, even though the conversion rates will be lower. -- the objective function can change. It can change it at any time. And it should change. It should really change depending on what the management of the business wants it to do. But what should be consistent is never interact with a customer without taking into account everything you know about the customer. So if you've had 25 transactions with the customer over the prior 24 months, how about we use that knowledge in what we do with that customer. So what we expect is when you show up, we know your brand proclivity. -- we know your price elasticity. We know what makes you tick. We know how to get you to respond to whatever it is we want you to do. That's what FICO of forward. And it does here better than anybody else's. There's nothing that comes close. So a lot of B2C companies have tried to build this themselves. Usually, we don't compete with other software companies. We compete with homegrown because there really are no solutions quite like ours on the market. And then we are so much lower cost than them building it themselves. They'll go to do something that's halfway is good, they'll spend $100 million or they can buy ours for $15 million. And so that's kind of the dynamic. So it's very popular. We're growing really fast.

So maybe just a follow-up there. Like you said, a lot of -- the way you described it, a lot of software companies, a lot of analyst companies try to say the same thing. [indiscernible] assets, brands within that, that help you keep growing at this rate .

Well, if you think about part of the reason that you don't see a lot of analytics software companies is really, really hard to put analytics into software. That's because what do analytics do? You ask a question, you're looking for an answer to a question, and then you have to figure out which data is the right data to answer that question. And then you have to figure out what's the right analytics technology to put on that data to get to the answer you want and do that all in real time. That's a very complex equation. I mean, it's just really hard to do. And we've mastered it over 40 years, and no 1 else has. .

Got it. And maybe to that point, AI, the new LLM, they're making life easier for you? Are they introducing more competitive threats? What's [indiscernible].

For us, it's good on all counts. So we get the same productivity benefits that any software company would get from no longer having the right code, right? So that's a labor saving. But more importantly, we think that all of the adjacency -- the core of our platform is decisioning. We think that all of the adjacencies lend themselves really well to AI-driven agentic behavior. So I don't know, let me give you an example. -- we try to detect fraud. And so we'll go in and look at millions and millions of millions of transactions and try to identify patterns and profiles and figure out what looks odd -- and when we find those things, we either -- if it looks okay, within 17 milliseconds, we say, looks good. And then the rest of the time, it's an exception gets kicked out to a fraud analyst. Well, the fraud analyst takes that exception and studies it, it looks into your account and checks your checking account balance and other kinds of things, it makes a determination, not in 17 milliseconds, but maybe in a half hour. In our view, an AI agent could do what that analyst does in another 30 seconds. So instead of 17 milliseconds, we get the exception kicks it out, and now we have an AI agent that does what the fraud analysts used to do. So I think that the banks are going to get tremendous savings out of applying agents to the outcome output of our software.

Got it. And maybe the last question in the last 2 minutes here. It sounds like you can really need decision platform. You've been accelerating growth recently, doing very well. What is the vision for software you? How do you -- it's still relatively small in the schema of software. So how do you scale it to the heights that you probably want to take this? .

Well, I think if we keep growing the new platform at over 40% a quarter, it's it will scale to new heights very quickly. It has been growing really fast. I mean the -- it's now -- the new platform is 1/3 of our total software business and growing much faster than the rest. The rest of it is pretty much flat. And so we've got the growth. And we'll see. I mean, I think -- and also our model is a land-and-expand model. So anything we land, we typically get a lot more revenue out beyond that. our dollar-based net retention revenue on the platform is 136%. So you can see that the customers who buy it, buy more. I mean that's kind of what that means. We'll see. We don't have -- every time once in a while, I get asked questions, what are you going to sell it or spin it off. And of course, that's a question the Board asked once a year. But we have no intention of doing that anytime soon. Right now, software is not in favor -- this wouldn't be a great time for us to spin it off anyway, but we really think it's undervalued and underrecognized.

Okay. Cool. All right. We'll leave it right there.

Thank you.

Thank you.