ExlService Holdings, Inc. (EXLS) Earnings Call Transcript & Summary

Good afternoon. My name is Puneet. I'm from JPMorgan's Payment Processing and IT services team. Glad to have here with us Maurizio, CFO of EXL. You all know him well. Last presentation of the day, but saving best for the last. So Maurizio, thanks for joining us. Format of this presentation is going to be fireside chat. I'll start with a few questions, and then we'll open up the floor from questions from audience. So thank you again, Maurizio.

So maybe for benefit of investors who may be new to the story, could you talk about EXLS' positioning, your value proposition and how you are unique relative to your peers?

Sure. And thank you for having me, Puneet. So why don't we talk a little bit about EXL's value proposition? When we think of the value that we propose to our clients, it's really in three different areas that are core to the overall business. The first is domain expertise, in that we understand the segments of our clients' workflows significantly because we've been at operations since essentially the early 2000s. We manage many, many workflows for many different of our clients within different industries, whether it's insurance, health care, banking, many different areas. We've built that domain expertise over many, many years. Secondly, data management. Back in 2006, we made the acquisition of a small Indian company called Inductis, and we got into data analytics. And since then, we have built up our data management business. And now that's becoming very significant for us, especially in this AI world. Everything starts -- when you start to build AI, it's on client data. And in many cases, we need to work on the client's data in order to make it ready and structured in order to build AI on top of. And if you notice, about a month, 1.5 months ago, we had a press release that we came out with EXLdata.ai. So what is that? That is 65 agentic AI agents that we use, and it's an accelerator to be able to use that functionality to structure and manage the clients' data to get it AI ready at the end of the day. And so that data management piece is really core overall to our proposition. And then lastly, it's us building AI solutions for our clients. Now those can be AI solutions that we sell in a proprietary mode or it's AI solutions that we embed into existing or new client operations that we bring onboard into our overall workflow. So between domain expertise, data management and AI capabilities, when you put all three of those together, that is an extremely valuable proposition to our clients because many clients will -- or many competitors will have one but not the other two, right? They may have domain expertise because they run operations, but they may lack the level of capability in data management. Or AI or it could be the other way around. It could be an AI-specific business that builds AI solutions, but does not do much in data management and does not have domain expertise in certain industries.

No, very good. And clearly, like it reflects in the growth rates, you are clearly outperforming your peers, IT services companies. So let me -- so let's disaggregate data and AI revenue that you talk about. How much of that data and AI-led business is data-driven, so like you said, like helping clients improve quality of data, helping them access data? So is there a way to think about like how much of that data versus, let's say, legacy analytics work that you've been doing for decades now? And also, maybe if you can also talk about marketing analytics, which has been an anchor to that segment's growth in the past. Is that still a headwind?

Sure, sure. So our Data and AI segment is the piece of our business that is really growing significantly now. We are embedding data and/or AI in so many of our clients' workflows. Every new opportunity is starting to have some form of data or AI embedded in that new opportunity. We're also cannibalizing and going after our existing Digital Operations business and embedding data and AI into that business overall. When you look at the total, when we released our third quarter results, we now have 56% of our revenue as data and AI. So what's the makeup of that piece? The big piece of that is our historical data and analytics business. When we ended the year last year, about 44% of our business was data and analytics. So that is buried in that 56%. The other 12%-ish of that total is the remaining us selling our AI solutions into the marketplace. And so as we move forward, you should see data and AI continue to grow between 15% and 20% going forward. And as Rohit talked about on the call, digital operations is started -- we're starting to really embed more and more data and AI into that revenue. So you're going to see some of that revenue actually move to data and AI. So that growth rate will probably be somewhere in the mid-single digits going forward. When you aggregate it all together, you still get to low double-digit growth overall. And like this year, our organic growth rate will be right around 13% overall as a total consolidated growth rate, with data and AI being in the high double digits or high teens and Digital Operations being in the single digit.

Let's double-click on Digital Operations, like mid-single-digit growth that we expect there. Obviously, that segment also suffers from AI cannibalization, like that revenue moves into the other segment when you automate a process using AI, right? So the segment was down like 2% sequentially in the last quarter that drove like a lot of concerns among investors about the long-term growth of that segment that if AI cannibalization could keep that segment from growing at mid-single digits. So talk to us about your confidence in that mid-single-digit growth in that segment. And what will drive that incremental revenue to offset cannibalization?

Yes. So there's still plenty of opportunities in Digital Operations. Even though the majority of our Digital Operations business will have an element of data and/or AI in it, we are now -- and you're starting to see that in our results, whereby we are actively going into our Digital Operations business. And we used the example of our British energy client in the call in that we went in into one of their workflows, and we got a 30% efficiency from that overall workflow from embedding agentic AI into that process. And so essentially, the client got a 30% efficiency benefit. If you think about it in terms of revenue, the revenue comes slightly down because of that. But what ends up happening is because we've embedded AI into that business, we're getting more of the work from that client because we have to handle the workflow from end to end. And because of that, nothing -- we did not see any material change in revenue. What we did see was profitability increased materially for that client because now a big portion of that is technology driven. We also become a stronger strategic partner for the client because they've seen us embed AI into their workflow. Our success rate in embedding AI, particularly in operations, is 90% plus because we run the operation. And so we're able to embed AI successfully because we're already running that operation. So what you end up having is an extremely positive effect overall with the client. It's much better revenue for us because it becomes technology IP driven at the end of the day and it gives us the opportunity to expand within that client. And if you look at kind of the growth of our company over the last 4 or 5 years, and we have it in our investor deck, we continue to expand within existing clients, more workflows and increase in total average dollar amount that we're billing the clients because we're taking on more and more workflows. And that's been the way we have grown the company essentially since the beginning of time. It's really a land-and-expand approach.

Yes. So the new contracts that you're signing in Digital Ops, like I'm assuming like almost -- in almost all of them, there is some level of AI benefit that's baked in that. So contract comes up for renewal, when you renew, the client would expect some level of AI benefit if they're locking themselves in for 5 years, let's say. So are you seeing like in those cases, some level of overpromising by your peers, like overpromising AI benefits? Because you are -- like AI is still evolving, right? So do you off-promise benefits of what AI can do today or what you think AI will be 5 years from now? And so could that create like a scenario where some of your peers might -- who are hungrier for growth, who are finding it difficult to grow, overpromising what AI-based benefits?

I think clients are pretty astute, right? I think with the evolution of Gen AI and now Agentic AI, you have many, many more competitors out there in the market. We traditionally have competed against the traditional BPOs. We've traditionally competed against someone like Accenture and a few others. But now you're seeing consulting firms with their own AI practices. You're seeing some of the big 4 accounting firms with their own analytics AI practices. You're seeing a number of other types of players in the market trying to sell AI solutions. So when you -- when we go to -- when we start the process and bid for a piece of new business, you're seeing many competitors. That funnel really starts to shrink as we get further into the bidding process. And you end up and the client really understands what each of the capabilities are, who they're willing to trust, who has the domain expertise, data management and AI capability for all three to be able to implement. And that really narrows the field down to 1, 2 or 3 competitors when you get down to it. So I think clients are astute to it. And so I do think you start with many more competitors now, but the funnel comes down pretty significantly.

And do clients typically wait until like a contract renewal to renegotiate or to have like their vendor bake in AI gains? So if you -- let's say, if you signed something 2 years ago with 3 more years on that contract, would clients wait 3 more years? Or would they say, proactively go to the vendor and say, "Hey, let's figure out, let's see how you can include AI" and renegotiate the contract?

It really depends upon the client, right? So there's going to be clients whereby we implement AI and we determine to wait until the end of the contract. Some other clients, we will embed AI and we will repaper it and discuss the new terms on that contract. If we're embedding AI in a contract that is transaction-based, we -- that waits until the end of the contract because the client is just paying on a transaction basis, whether there's AI or not. But for us, it's important because embedding AI means we will have a higher profitability because there's less people involved. So it really depends on the client and the process and how it's getting built.

Let's talk about Agentic AI. Obviously, like a lot of hype, a lot of focus, energy there. How much of that is real in terms of like the actual use cases or actual dollars like where clients are replacing their current process with an Agentic solution? Like is it still like very, very early stage, more like POCs and pilots? Or are you seeing like that it is large enough that it's meaningful to move the needle for EXL?

I think it's getting more implemented. I think, in order to move the needle for us, it has to be pretty material, right? So we're not there yet. It's obviously, we're not there yet. But as we create more and more Agentic AI agents, you're seeing it have more an effect in different client operations. And as we go quarter after quarter, you're going to see kind of this effect going forward. We had our first quarter in the third quarter where you saw Digital Operations actually decline in total dollars, right, because we embedded AI, and a lot of that is Agentic AI. So you're seeing it start to move into client operations. Some clients are willing to implement it more quickly than others. And some are taking a little bit of a wait-and-see approach. But for us, we're using that technology not only to embed it into client operations, but also in other areas like data management, like EXLdata.ai is literally us creating 65 Agentic AI agents to use in transforming client data. So we're using the technology in many different areas. I think clients' reception is a little bit mixed. But there are clients that are moving forward with it, and we tried to highlight that in the call.

Yes. AI readiness has been like a topic that has come up in almost every of meetings today. So talk to us like what are the constraints in a client organization that's keeping them from embracing AI. It could be governance like the security, privacy, change management issues, it could be the data is not ready, like you talked about like you help clients transform data or it could be like the use cases like there is not enough conviction in those AI use cases, the benefits they might get, like if I go back to that MIT report that everyone talks about. So talk to us like when you see your clients, like what are the bigger constraints that's keeping them from embracing AI? And how can EXLS help those clients overcome those barriers?

Yes. I think it's a little bit of all of that. I think, one, it's security clearance overall. I think that's an impediment for a lot of the highly regulated segments that we operate in, whether it's insurance, healthcare, banking, I think that's an impediment and that delays the process overall. Data management is critical, right? So in most cases or in a lot of cases, the clients' data is not structured in a way to implement Agentic AI or an AI solution on top of. You got to keep in mind that some clients still have not moved their data from on-premise to the cloud, right? And the cloud came out 20 years ago. So clients are still moving -- doing that work and then looking to do work such as Agentic AI on top of that to be able to become that much more efficient. So I think there's a bit of -- still of a process that clients are going through. But there will be clients that will move very quickly to implement this technology. And that's why we want to ensure that we have the solutions for them to take advantage of that sooner than later.

At this time, are there any questions from audience?

I was curious if -- it's kind of like a higher-level macro question. If you just -- if you think about like your clients and how they're engaging with these types of projects, does it feel like there's a holdup around macro uncertainty still? Or has this been kind of like -- because I feel like we've been talking about this concept of uncertain -- macro uncertainty for like 3 years. Is that behind us now? Does it feel like there's still an unlock that could be kind of achieved if a macro turns one way or the other? Do you get the kind of crux of the question?

Yes. Yes. I think clients are looking to assess the ROI on these solutions. I think the macro has changed a little bit. If the macro gets better, I think they'll be a little bit more aggressive. I think if the macro gets a little bit weaker, maybe they slow it down. If they see the ROI on implementing that solution, I think they will move forward. I would say, probably 1.5 years ago, in the banking sector, we saw a little bit of a hold on budget spending. We don't see that today. We see our banking business doing very well today and a lot of our global banks implementing more of our AI solutions and doing more data management work with us. So if they are releasing their budget on the banking side, then I think the macro is actually not in a bad state today, right? So I think it's less macro. It's more about them willing to implement and seeing the ROI so they can move forward and getting over that hurdle that we talked about just with -- in terms of impediments.

My question is about ROI on like AI and agentic solutions. Everybody is talking about the MIT report this year, and we've heard a lot that a lot of the preliminary POCs didn't really get off the ground, didn't generate a lot of tangible ROI. So I think maybe the question -- the way I want to ask this question is like what are people doing wrong if they're not achieving results? And like what have you seen in the market where people have made misplaced investments on AI that didn't really pan out versus what you all are up to?

I think we've been a little bit more successful with POCs because we're doing POCs in segments we operate in. And so we understand the process a bit better. It also points towards the success rate that we have in implementing AI -- so we have been -- we've had a higher success rate with POCs. And a lot of those POCs have made it into production now going forward. So I think we've had that benefit. If you're an AI firm building AI solutions but you don't have that domain expertise or you don't have that data management capability, your success rate, from everything that I've read, is around 50% overall. That's a dramatic difference from our success rate. And I think that gives us a very big leg up. And I think that contributes to the difference on the success rate on POCs between us and maybe the rest of the market that has a little bit less of the three capabilities that we have.

You mentioned that you like to implement AI in your projects because from [ EXL ]standpoint, it's accretive to you guys. What's the pushback you get from clients in implementing AI in those projects? And is there a specific sector where you're seeing more pushback than others?

I would say there are clients that want us to take over an operation. And they first want us to take over the operation and then subsequently, at some point, embed AI into that process. They don't want to do it right away. And I've seen that in a number of international clients that we brought onboard, whereby we take over the operation and the discussion with the client is we're going to embed AI transformation in years 2 or 3 later on. So that still occurs. Now that means -- that also means it's good for us because we want that Digital Operations revenue also, right? Because we know at some point, that's going to get converted, right? And so once we have our tentacles into that operation, then we can embed that AI, but they're not doing it right away. And they're not doing it for multiple different reasons, right? It's kind of the stuff we talked about, security, they're worried about their data being unstructured or not structured well to build AI on or just a reluctance internally to do everything -- to start everything all at once.

So let's take a step back like and focus on the entire industry, BPO industry rather than EXLS. So let's say, AI adoption picks up. And what that does is it reduces the price like, let's call, prices as price per unit transaction, right? Not the bill rates, but the price goes down because you're doing services, providing services using AI. On the other hand, like the bullish we would be the quantity, the volume of work that will also increase to offset that reduction in price. Where will that increased volume come from? like -- because in BPO, like whatever the new volume, the new work that you will get, someone else is doing that work right now, right? So for the overall industry, I can clearly see EXLS gaining share, taking share in that. But for the overall industry, where will that increased volume come from? Who's going -- who's providing that?

I think when you look at just the overall landscape, right, so think about our contracts historically. We're always guaranteeing a 3% to 5% efficiency every year to our clients in our contracts, right? So that still remains, even in this new era. But as we move forward -- and we have grown off that, right? So when the contract comes up, there's a level of discount we'll give to the client, but then we get new work. Historically, that new work comes from client-run operations. And when we look at kind of where we continue to grow, there's still a huge opportunity in terms of managing more and more of our client-run operations. I don't think we've gotten to the point where any one segment is completely penetrated. Even if you look at insurance in terms of what can be outsourced, and I think only about 30% of what can be outsourced has been outsourced, and that's our most significant segment. So I think the opportunity still exists materially in the marketplace, in the different segments that we can take on more and more of client-run operations. Rohit talked about it on the call that the opportunity just in healthcare alone is enormous for us. We are extremely small in that area, particularly in operations. So that is just an example of many different areas within in our client set that we can grow into just gaining client run operations, forgetting competing with -- for operations that are run by our peers.

And are you seeing -- like you talked about earlier, like that you're seeing not just companies like Accenture, but also big 4 consulting companies trying to compete for the same opportunity. So are you seeing like -- how should we think about BPO companies' competitiveness? Like obviously, you have the domain knowledge, you have the process expertise versus some of the IT services companies, which can bring like the technology capabilities they can help move -- help clients move to cloud, improve data and all that. So how do we assess like these two priorities or focus that BPO versus IT services companies bring to address this opportunity?

Look, I think all of our peers are trying to poke into different directions into each of our businesses, right, or our business models. I think, for us, we are very well positioned in terms of many different areas. One, I think we're well positioned in Digital Operations with our domain expertise. I think that is core to us really continuing to build out the business. You're going to try -- you're going to see some of the peers or like the IT companies look to get into that space. And they've made -- some of them have made comments to try and do that organically. But again, it also comes down to who's the client going to be most comfortable with at the end of the day. And a lot of these companies don't have the other capabilities that we have, particularly in data management. I think everyone is trying to build their AI services or capability. But I think the domain expertise and the data management capability are a bit unique. And they -- you can't just build that over a 12-month period, right? That is years of knowledge. And to a certain extent, it's almost like an IP to us.

So given like all those changes that are happening in your delivery structure model, how should we think about margins? Like in slight margin expansion at EBIT margin level is still the right goal?

So when we look at the trajectory of the business, right, and you look at 2026, we should be continually driving data and AI, the percentage overall. That should continue to drive higher the percentage of the overall business in data and AI. As we do that, that should drive overall gross margin, right? We talk about driving gross margin so that we can invest more below gross margin. The net of all that should be an increase overall to adjusted margins, right? And we do that every year. Now some years in the past, we've done a lot of increase in adjusted margins. But now you're going to see us invest a bit more in R&D. So we're going to continue to grow EPS faster than revenues. If you look at just at 2025, EPS growth is between 14% and 16% in our guidance and revenue growth is 13%. We have a mandate internally to drive EPS faster in revenues. In order to do so, we have to drive margin. So you're going to continue to see margins increase on an annual basis. They may not be as high as you saw in '21 or '22, whereby in some years, we've increased it more than 100 basis points; but that's still going to be a goal of ours on an annual basis and a target.

And can you also quickly talk about use of cash priorities? You haven't done an acquisition in -- it's been almost a year, I think. So talk to us like how do you see -- like with valuation, especially with valuation for IT services or for some of those technology capabilities, public valuation at least coming down so much, how do you see like use of cash in terms of acquiring a strategic asset tuck-in deal versus other uses?

We're still very inquisitive on M&A. We still look at a lot of assets, I think. And the assets we look at are particularly in data and AI, capabilities in data management and also digital AI capabilities also are -- that we can build out horizontally in all our segments. It's very attractive to us. So still very inquisitive. For us, it's more tuck-in acquisitions or medium-sized acquisitions that we're most interested in. And if we're not allocating capital to M&A, then we're buying back shares, particularly in this environment. When you look at the stock price today, that's very attractive to us in terms of our buyback. We were authorized for $0.5 billion in buybacks back in March of 2024. That will expire in March of '26. I think we did over $200 million last year. We'll do over $200 million this year. So we'll continue to buy back shares at this price because when you look at the intrinsic value of the company, looking at the future profitability and cash flows, it's an attractive price today on where it is. And for us, the buyback shares today is a good use of -- and we believe it is a good use of our cash.

That's amazing. All right. Thank you so much.