TELUS International (Cda) Inc. (TIXT) Earnings Call Transcript & Summary

All right. I think we can get started. This is the session for TELUS International. My name is Tien-Tsin Huang. I follow the IT services and payment space at JPMorgan. And always excited to have Jeff Puritt here, President and CEO. Jeff is always a great person to talk to, and I'll always learn something. And he's got great enthusiasm for the sector, for the company. So Jeff, thank you for being here.

Nice to be here.

We have a lot of questions to get through that I put together with the help of investors. But I thought, just for the benefit of those on the webcast and those in the room that aren't as familiar, just maybe a quick commercial on TI and your right to win, and then we can get into details, if that's okay?

Sure. I'll try not to go on too long. We're a technology services firm that brings together the best of talent and technology to design, build and deliver innovative solutions for global and disruptive brands. We've been around for 19 years now. We've achieved some meaningful level of scale, about 75,000 full-time team members, supplemented by a little over 1 million members of our AI crowd community. We deliver our solutions principally from 32 countries, 68 delivery centers and design studios around the world. Focused principally on 5 sectors: tech and games, telecom and media, e-commerce and fintech, BFSI and health care. Dabbled a little bit in a couple of others, like retail and travel and hospitality, but to a lesser extent. Really focused on 4 service lines: AI data solutions; trust and safety, which includes content moderation; client experience management or CXM; and traditional digital IT. We've spent a fair bit of time expanding our capabilities in terms of reach, depth and breadth, both organically and, more recently, in the last 6 years, inorganically, making some pretty meaningful acquisitions and integrating that into our core offering. Everybody is talking about GenAI a lot these days. And we've had the privilege of serving many of the hyperscalers, not just for a long time in traditional machine learning algorithm, data annotation collection and relevance, but pivoting over the last 3 years, in particular, to supervise fine-tuning and prompt engineering to help many of the hyperscalers build their large language models, and now using that expertise and capability to help enterprises take advantage of these capabilities now. And I suspect we'll talk some more about AI or generative AI more particularly, and we can talk about our Fuel iX offering specifically. We had a tough 2023, as I think a lot of folks in our sector did, but managed to recover through the back half of the year with some solid growth and return to, I think, near best-in-class profitability and set ourselves up for what we hope is going to be a better 2024. Obviously, still mindful of the narrative around the macro. And a lot of folks talking about the back half of the year, maybe it will come back a little bit, maybe it won't. We've made some specific incremental investments, well above levels that we've historically done in terms of sales and marketing, to try and will take advantage of and to stimulate more opportunities around growth, not exclusively, but again, around GenAI-enabled capabilities in particular. And then, obviously, drinking our own champagne and eating our own gourmet cooking, as you've heard me say often, transforming our own cost to serve, leveraging these very same tools and technologies and process excellence and reengineering so that we can continue to generate these industry-leading margins and ensure the headroom in our cost structure so that we can address the constant pressure on pricing coming from the market more broadly.

Yes. No, it's a great summary. I think we're done.

All right. I'll see you at the bar.

So a lot to talk about on that great intro, Jeff. But I've been asking a lot of the companies, and we're so thankful to have all these IT services companies come, including yours. We've been dealing with this demand pressure basically since the second half of '22. I've been doing this for probably a really long time, call it, 24 years, but I don't remember such a long down cycle. So I'm just trying to understand and ask everyone why that might be the case. Is this purely just a cyclical issue? Is there a hangover from the pandemic boom of spend? Or is there something else that's structural or secular?

So speaking of hangovers, I think my hangover is recovering [ well ], but force me to give you a 2-ended answer. And I think it's both cyclical and secular. I think from a cyclical perspective, there's no question that the macro, whether it's persistent interest rate inflation, concerns around geopolitical stability, supply chain, the list goes on and on. I think those are persistent for sure. And I think you're exactly right, back half of 2022. I think there was this euphoria coming out of COVID and virtualization everywhere, a lot of stimulated rapid growth and hiring and spending and all the rest of that. And all of a sudden, things, I think, got pretty tight and decision-making got constrained and/or elongated in terms of the cycle. From a secular perspective, I think AI is really creating a significant disruption, both in terms of curiosity but also in terms of paralysis. Everyone recognizes the, I think, nontrivial disruptive impact that AI, more broadly, and generative AI, more specifically, not will have, but frankly, is already having on almost every industry. And I think folks are quite curious about how they may take advantage of that and are equally concerned about what the negative impacts on their business, on their customer relationships might be. And of course, a lot of us are out there trying to persuade businesses that we have just what the doctor ordered. And there's obviously some trepidation, concern, interest, as I said before, curiosity about who is the right partner, what sherpa should you select to help you on this journey in leveraging the benefits of this disruptive technology. And I think it's not entirely different than whether it was RPA or digital transformation, more generically, more broadly. Even just 5, 6 years ago, everybody recognize you got to do something, but what and when and how and with who. And so I think it's going to be several months, maybe even another year or 2, before we really see the implications of the have and have not as a result of this disruption.

Right. But you're running towards it, right? I mean, as they say, you're running towards the roar. And I think you mentioned you're trying to stimulate demand through some incremental investments, including across GenAI. I mean, you mentioned Fuel iX. So maybe let's just talk about it now, like what's the game plan here to stimulate that demand in that context of understanding you got to run towards this opportunity?

Yes. I mean, I think the near-term gold rush for us has been helping the hyperscalers build these large language models, and we've been privileged to support a number of the large language model builders, but particularly Google. As you will have seen, we were up 22% in terms of Google spend. Now our second largest client by revenue last year, '23 over '22. Q1 2024 over Q1 2023, spend from Google, up again another 30%. And although we support Google across more than a dozen, I think we're about 17 different service lines and programs today, the lion's share of their spend with us right now is in AI. We kind of see this supply demand bifurcation around AI and GenAI, in particular. And the gold rush today is helping to build these LLMs. But for tomorrow, literally tomorrow and for arguably the foreseeable future, where we think the real growth opportunity around GenAI is helping enterprises in the mid-market to take advantage of the capabilities that are represented in this pretty exciting technology. And what has already sort of revealed itself in these early days has been sort of the 2 Achilles heel, if you will, of the LLM environment is the hallucinations, which is a poly word for just plain old getting it wrong. And I don't think there's too many industries that can afford to get it wrong, whether it's health care and financial services, in particular, but frankly, nobody wants to be on the receiving end of those kinds of hallucinations. And then equally importantly, is the loss of control over proprietary private confidential information. And it's not just a copyright infringement, although that's a concern as well. But just generally, if your business with unique proprietary technology, like Samsung's developers did when they're using the ChatGPT engine to try and help them optimize their code, and before they knew it, their code is now in the public domain. I think the future of using generative AI is to find a way to mitigate the hallucinations and protect against the loss of control and privacy. And so leveraging our expertise, working with the hyperscalers to now bring the value of generative AI into the enterprise environment on a private cloud environment and really leveraging RAG, or retrievable augmented generations, you kind of dip your toe into discrete databases on a pull rather than on a push. So you don't lose control because you're using the proprietary data of that enterprise, you improve the likelihood of accuracy each and every time your GenAI model is producing an outcome. All of a sudden, I think this becomes a more robust and reliable ecosystem. And so down the road, how we're positioning Fuel iX is sort of a generic orchestration, integration and management layer called Fuel iX Core, over which you can lay either our own proprietary tool sets, whether it's agent assist or summarization, translation, et cetera, or open API architecture that allows you to use somebody else's insert name of application here, any of the ones I just described or more. And so in the fullness of time, we kind of see the opportunity to give enterprises access to this pretty cool technology without the kind of lock in to a single ecosystem, the same sort of challenge that I think folks just a few years ago when they wanted to take advantage of going to the cloud, got nervous about when they put all their virtual machines and data in one hyperscaler's cloud, and then they get locked in to that pricing model and that tool set. In order to give themselves a little bit more optionality, they wanted to split it up. So too, around generative AI, we see ourselves as technology-agnostic, enabling access to all of the best-in-class tools.

Perfect. No, I think that's a great description. With Google, people don't -- people think of as strong engineering, and they do everything internally. And the fact that they're engaging with TI, I think, says a lot. So just quickly for everyone's benefit, how did that relationship develop? And how did you earn that trust?

So we've been an enabler for Google, gosh, since 2011. And we started doing technical support for them when they first launched Google Apps for Work, some of their competitive productivity suite of tools looking as an alternative to Microsoft Office. So Sheets and Slides, et cetera, instead of Excel and PowerPoint. And from then till now, our relationship has really proliferated on the back of delivering reliable, quality, timely, competitively priced services. And one day, maybe that will change. But as I think I shared, we had a pretty good '23 over 2022 with them, and our second largest client by spend, and a terrific Q1 with them. We're already seeing automation around AI. So ourselves, we've built some proprietary tools, whether it's our expert assist engine, which algorithmically matches the tasks that Google or others have of us with the community that are required to do the supervise fine-tuning at scale and speed as well as more traditional annotation supporting Google Search Optimization, Maps, et cetera. In fact, winning spend from competitors of ours who have been in the news of late and Google bringing that spend to us as well. The automation we're doing now with our fine-tuning studios is to really try and continue to focus on expertise at scale around STEM disciplines, as well as health care and a few others, where there really seems to be an opportunity to maximize the accuracy and reliability of these engines, capability of generating reliable, accurate, truthful responses. Automation is here now and spend is not diminishing. Does that mean it will be like that forever? I don't know. But thus far, we think that what happened with the search algorithms in traditional machine learning was you build the model. But if you want the model to continue to be relevant and accurate, it needs to be updated and optimized all the time. And there are hundreds of every year incarnations of those updates. We anticipate the supervised to fine-tuning activity will similarly require a near-perpetual revision in order to ensure it's always relevant and up to date. So for now, it feels like we're in the right space. If it changes, I guess we'll take you one day at a time. And part of our supercharging our sales and marketing spend is not just specific to Google, but frankly, one of our other large hyperscaler clients, where their spend with us was rather diminished in '23 over '22. That client concentration profile comes at a cost. And so trying to as quickly as we can proliferate, derisk that profile, win new clients and then land and expand in an effort to try and reduce the impact when one of them decides to spend a little bit less with us as part of the challenge and the opportunity we're pursuing right now.

Okay. Yes. So before I dig into that, just with what you just laid out with automation, Jeff, and from a modeling the business and you're building out the P&L. How does that change the equation on whether it be revenue per FTE or productivity on the revenue side? How should that generally translate from a learning standpoint?

So admittedly, it's still really early days, less engine, and I don't think it's true objectively empirically. I don't think it's true that our revenues in 2023 or certainly even in year-to-date 2024 have been palpably affected either adversely or favorably by GenAI, either inside TI of our own accord or by competitors, by customer preferences. But we're anticipating, and as we build out our own models and think about how we need to restructure and pivot, which is a word that, I guess, is also almost as overused as AI these days. If it used to take $0.60 worth of labor in order to produce a $1 of revenue, we think, prospectively, it's probably going to be $0.40. And we're anticipating, and we need to inoculate ourselves against the likelihood of circa 20% to 40% of productivity impacts coming from pervasive deployment of generative AI. So for ourselves, that means we need to, as quickly as we can, deploy generative AI-enabled tools to both support our customers and self-serve so that our customers end user community can get what they want from their customers online as seamlessly, effortlessly as possible. And if/when those self-serve channels are insufficient, then the next line of defense or offense is to arm the humans in the loop, again, with supercharged data-intensive insights that ensures that, that interaction between an end user and a representative of a business can be as effective and seamless and expedient as possible. And that means, again, indexing the entirety of the data community inside that business. Having generative AI automatically populate the agent desktop with an understanding of exactly who that customer is, what they were buying, what questions they asked the bot that couldn't be answered, and make a recommendation, next best answer, rate then and there. All of that should reduce the amount of labor that goes into generating the dollar revenue and should, in the fullness of time at scale, generate better margin return. But these are early days. And so, again, we're deploying pilots all over the place. But it's hard to charge full freight for a pilot when you're trying to prove out your credentials and capabilities and you're experimenting at a relatively small scale in order to avoid potential catastrophe if you just deploy everything all at once into your customers' end user community. And the pilots we've run with a number of financial institutions in the U.S. and Canada already, not surprisingly, they're not putting these GenAI-enabled solutions for their customer community to find out, oops, that was a big hallucination. And all of a sudden, my entire savings is gone, or the recommendations are misaligned with reality.

Yes. Okay. Yes, so very much in discovery mode, and you'll learn from there. You mentioned it briefly, so let's ask -- let me ask about the stabilization with that social media client. And are some of the efforts that you're talking about simulating growth apply to that account at this point? And just a broader question on content moderation on top of that.

Yes. On the former, so disheartening, frustrating not seeing the kind of growth potential that, frankly, we could have or should have, and that's on me, that's on us. 2023, it was a tough, bad year in terms of the year-over-year compare and in terms of spend derived from serving that client. I think we were unnecessarily, unduly circumscribed in terms of near-perfect single threading service line and geography. And so when the decision was taken to meaningfully curtail that spend, we didn't do much to help ourselves in terms of mitigating the impact that, that had. Europe is a tough labor economy to shed surplus labor in a timely fashion. Layer on to that and the not just the regulatory legislative regime that exists in Germany and Spain, in particular, for example. Then we have workers' councils that populate our businesses, then that adds another layer of complexity and time and delay in expense in order to try and rightsize your labor force with your volume demand. And that was problematic for us. Thankfully, we've chunked through most of that. And then, more excitingly, into Q1 of 2024, we're starting to see a rebound in some of the demand. There is still a lot of work for us to do in terms of further derisking that relationship, serving them from [ geo ] to through end, serving them long service lines to through end. But early days and lots of green shoots. So even if I can just stabilize the decline and not lose as much business in '24 as they did in '23, that will have a meaningfully favorable impact on our outlook for the year. And then in terms of content moderation more broadly, like 95% of the content that is moderated today is already moderated at least at first instance by an AI-enabled tool. The residual that requires humans or humanity in the loop is sub-5% of the overall volume. But interestingly, our species predilection to populate content to the web is seemingly limitless. And the growth rate continues to be geometric, and our creativity seemingly has little to no bounds. And so these AI engines, even when we're talking about generative AI, and we're not quite there yet in terms of seeing generative AI being used to moderate content, it's still legacy machine learning algorithms that are being relied upon. By definition, they rely on history to inform the future, to identify what comes next and to accurately characterize it as either objectionable or incompatible if you're talking about ad placement, et cetera. And that's where humans in the loop come in. So when humans are being ever more creative, use algo speak by way of an example, so your interest for some different characters with traditional [indiscernible] in order to pass along a different meaning. Hey, I can always pick that up. Never mind slang and context, et cetera. And never mind, in fact, that most of these engines are principally English-centric. So for other languages, whether it's Middle Eastern languages or European languages, not yet quite the same robust technology-enabled automated solutions at scale. So we think there's going to be a viable opportunity to continue to have humanity in the loop doing content moderation, our digital-first responders as we call them, for a long time to come. But like so many other things, again, you got to kind of move up the food chain. So you need more talented human beings who can interoperate with technology on a real-time basis as those filters that are AI-enabled, look after sort of the lion's share of the content, but I'm not sure it gets much past 95%, 96%. And there's still lots and lots just in terms of volume of content even in that 4% to 5% range.

Okay. Now I know you started out being hard on yourself on the account, but it sounds like it's just geo and service line exposure and stabilization there, content moderation use went through, and the hope is that there's more work to do still as you just diversify. Does that take long?

Yes. I mean. I could curl up in the fetal position in the corner and rock myself to sleep to try and feel better, but I'm not sure it's going to get the job done or give us a better outcome. The reality is there is opportunity for us to do more, and we're at it. But 80% of a good strategy is execution, and we got to get after that.

Okay. [ English cost ]. So on the good side, the parent company, TELUS, has done very well, and I think that's something you have been very bullish on, Jeff, and exposing them to WillowTree as well. And it sounds like that's in a good place. So what more can you do? And when I talk to paint about the name, it feels like that's a great case study on how you can service a large multinational MNC to do a lot and expose a lot of your service. So tell us what -- how much more is there to go there? And how replicable is that client to other players?

Yes. I mean, in a perfect world, every one of my 650 clients would be just like my relationship with TELUS, where we serve them with 300 different programs, almost the entirety of our solution set, and we're growing the business 20%, 30% year-over-year. Unfortunately, we're not there yet. The plan, of course, is to take what we do within for TELUS and repurpose that success for not just other comms and media clients that look, act a lot like TELUS, but even in other industry verticals. But specific to the growth potential within TELUS, I mean, for Q1, again, the growth is pretty darn exciting, double digits. For full year 2024, we anticipate, again, it will be double-digit growth throughout the year, and we have pretty strong visibility to that. I think what makes TELUS a little bit different than a lot of other businesses, particularly on the telcos, is TELUS, as you know, has already made the decision to diversify away from just pure-play telecom. So their health care business grew, for us, in terms of our health care-specific revenues, 173% last year year-over-year, admittedly off a relatively small base. So the numbers flattered somewhat. But first quarter year-to-date, up another 23%, and there is a ton more of growth as the health care business in TELUS grows. They continue to rely upon TELUS International and our broad range of services to enable that capability. We're still just scratching the surface on that one. The $2.7 billion acquisition of LifeWorks that TELUS made a little over a year ago, and we're just still enabling the realization of synergies on that front and then supporting their growth. There's lots of upside to come. And wait, there's more, TELUS Agriculture & Consumer Goods, again, a relatively nascent portfolio business inside TELUS, where we're already doing some enabling work. And as they grow, so, too, will we. So I think conservatively, there is double-digit growth to be had supporting TELUS as a result of its unique circumstances for several years to go yet. Of course, already, we enjoy, I think, a unique relationship with TELUS, not just because of the ownership structure, but as you know, we have a long-term master services agreement with a minimum spend commitment. And again, to the extent that there's concerns about concentration risk, I would suggest that they are completely mitigated in the TELUS context. We don't have those kinds of agreements or relationship with others, but we aspire to, and that's the job at hand is to try and replicate what we've done within for TELUS for other clients. And even at Google, where spend is bumping up against near-TELUS levels. We don't have an MSA, although it will be nice if we got to that point, perhaps. And then not just in the other hyperscalers, but across enterprise more broadly. That is the opportunity that exists. And again, I think we have a unique collection of assets in terms of capabilities and reputation and references. We got to do a better job of telling that story and showing the results of those capabilities. And then hopefully, before the coming of the next Messiah, our share price will better approximate the value that I think exists in this business.

Yes. No, again, anything is possible. But on the TELUS, are you using the WillowTree one as an example? I hear you on the new growth around health care and ag. But are you displacing incumbents in some of those examples? So tell us why I think it's a good test for WillowTree and its capabilities versus whatever incumbent might be there. What's going on there?

Yes. I mean, it's not unique to WillowTree. Years ago, when we started at TI, building up our IT capabilities, for example, we were displacing a lot of traditional IT providers doing back-office support at TELUS. When we built up a QA and QE capability, we were displacing QA/QE providers at TELUS. As we're building up sort of front-office design capabilities through the WillowTree acquisition, we're displacing those incumbents. So we displaced IBM and Infosys and Wipro and Tech Mahindra and Accenture and Thoughtworks and Globant and the list goes on and on. And as we build our capabilities, either organically or through acquisition, part of the opportunity has been -- will then go into TELUS. And I mean, I know there's perhaps some constituency out there that thinks, oh, boy, and you both have TELUS in the names. This must be a gift with purchase, just pitch up, and you get the work. To the contrary, we actually respond to RFPs from TELUS. I often argue, I'm not sure that's the best value for TELUS shareholders that a subsidiary, in part, has to respond to an RFP. But my colleagues insist that we do so in order to satisfy themselves that we are qualified, that we earn the right to win that business. And so it's both service quality and value for money. And revenue candidate keeps us all honest, if a parent company decided to start overpaying for services from an offshore subsidiary, there will be a tax implication. So we are benchmarked on our pricing and our service quality and productivity regularly to ensure that they're not running a fallow on any of that. So among the many things I'm proud of at TI is when we displaced Accenture from the QA/QE work or Thoughtworks from renovating telus.com, the list goes on and on. It's because we earned our way into those opportunities.

Right. Nothing is -- it's earned, not given. I think I've heard you say on this one before. Okay. No, that's helpful to hear because it is an important account, and we're trying to pay close attention to the larger players. I think we've covered the big 3. So anything else, Jeff, that helps inform your outlook for the second half and some of the improvement? I know a lot of it is execution around the stimulus, some of the growth. Any other signals that you're looking for that we should be looking for?

Yes. I mean, obviously, I'm sensitive to the broader narrative out there about the back half of 2024 and whether or not growth is returning to normal levels, whatever that means, at this point. I've seen and heard recently, even yesterday, some of my peers up here talking about their perception as to whether or not growth is coming back or not, whether back half '24 will look a lot like back half '23. I think, not surprisingly, there are different perspectives because none of us has a crystal ball. Woo that I did, I might not be in this business. I might be at the racetrack. My optimism comes perhaps not just from my own personality, but we had a pretty good Q1 in terms of bookings. We don't talk yet a lot about public disclosure around bookings. But it may just be a byproduct. We have more feet on the street. Because as part of that investment in sales and marketing, I've got more people out there banging on doors, et cetera. But we had a pretty good quarter on bookings. Now we need to flow through those bookings into profitable revenue as we deploy team members in connection with those projects that we won. And our funnel continues to be very, very healthy. Now admittedly, the funnel has stayed healthy, even last year was healthy, but decision-making kept getting delayed and elongated, so it didn't convert into end-year revenue as quickly as we would have liked. But I'm cautiously optimistic that the things we're selling and the conversations our teams are having, thus far, hopefully, will indeed manifest themselves in a stronger Q3, Q4 as we guided at the beginning of the year. And that, in conjunction with a less bad deterioration from customer #3 is, again, we're seeing right now, should in totality manifest itself in our ability to deliver against our guidance of 3% to 5% revenue growth, which is, in and of itself, not anything to write home about. But relative to my peer group, that looks like pretty d*** good.

Yes. No, it is. And like I said, it's a shame on us. We're always trying to force looking at comps and infer who's going to be right or wrong. But yes, like I said, it sounds like you have the bookings in place and some of the investments in place there. So is the bookings that you've seen, the strength in the first quarter, is that broad-based? Is it more vertical-specific? Jeff, anything else you can expand upon with the bookings?

Certainly more AI-centric, which, in and of itself, is encouraging given where, I think, the business is going more broadly as we just discussed. But it also comes with a bit of margin pressure because, again, as I mentioned earlier, we don't get to charge full freight on the proof-of-concept pilot. And so we need to be, as ever, and this too is not new, but I think it's intensified right now is finding that elusive balance of revenue growth and margin yield. I suppose, if all I cared about was top line growth, that I could tank the profitability of the company and give away our services for half the price. And as long as everybody was happy at 10% EBITDA, our top line growth would be pretty d*** impressive. But in the fullness of time, I'm not sure that's the right recipe for sustained success. And that's part of the challenge, I guess. And it's not like it's lost on the -- not like I was forced into accessing the public markets. And when you come here, one of the price of admission is, that there is an expectation in terms of quarterly results. So it is a constant battle to try and keep things sort of in the middle of the fairway in terms of top line revenue growth and profitability. And right now, it seems like I'm making nobody happy in terms of share price, but hopefully, we are here for the long term. And in the fullness of time, I think our strategy and the execution against that will bear out that we are delivering sustainable, meaningful value.

Yes. And you have protected the bottom line through and through, right? I think that's definitely been there, and we're all eager to see how the revenue comes through. But Jeff, I know the shares have been under pressure. I know the group has been under pressure as we led with this conversation. But what, in your mind, as you've been meeting with investors today and in the last several weeks, I'm sure, what do you think is really misunderstood that you're eager to sort of put the results up and put some of those things to rest?

I'm not sure it's particularly helpful for me to blame the market for where my share price is. At the end of the day, I think it's my responsibility to better articulate the value proposition to the market. And this is a long-term investment. I think we were fortunate, through our IPO roadshow, to attract some well-heeled, well-pedigreed, long-only institutional investors and not a whole bunch of day-trading hedge funds and retail investors. But there's a consequence to that in terms of how the shares are traded. I think our capital table has a nontrivial impact on what's happening with our share price, whether it's the control from TELUS, it's the overhang from Baring, the list goes on and on. And so whilst it might inoculate us to some extent from some of the variability, I think the net effect of having a really skinny public float means that the share price doesn't necessarily fully reflect the value that we have in the business. But in the fullness of time, my aspiration, my ambition is that the market will recognize that this is a business that is here for the long term, generating, right now, 16% free cash flow yield. And I get that tech services is not the space where investors generally go for value. They go to the dividend-yielding community for that. But it sure, I would hope, that in the fullness of time, a profitable business is always attractive and on strategy. And so if we can get back to meaningful top line growth and continued best-in-class margin free cash flow yield, I'm hopeful that the share price will recover and respond appropriately.

Yes. No, it's a long game. We can't obsess with the score in the early games. So I know you're working hard, Jeff. I always appreciate your time, and we'll be watching for sure for the next several quarters.

Thanks very much, Tien-Tsin. Appreciate your support.

Thanks, everyone.