TaskUs, Inc. (TASK) Earnings Call Transcript & Summary

I'm Ryan Potter, analyst from Citi's payments, processors and IT services team. I'm glad to have TaskUs for the next session. And from TaskUs, I have CEO (sic) [ CFO ], Balaji Sekar; and Head of IR, Alan Katz. Thanks for joining us guys.

Thanks, Ryan.

Thanks, Ryan.

Yes. So for anyone who's new to the story, I wonder if you could give like a 2-minute [indiscernible] pitch on who TaskUs is, how you guys have kind of evolved over recent years and how you're differentiated in the market.

Sure. So I'll take a few minutes. So TaskUs is a top-tier BPO. So we have 3 service lines. One is digital customer experience. Second is trust and safety, which also includes content security. And then, we have AI services where we do work like data annotation, data tagging. Like a brief history, we're going to be celebrating our 15 years this month. And so we grew up in Silicon Valley and working with some of the fast-growing, high-technology companies. And companies like if you open up for your phone, some of the apps that you see are the companies that are part of our portfolio. And then recently, in the -- probably in the past 18 months, we've been focused in terms of expanding from a geographic footprint perspective and also going after some of the large enterprise clients has been a focus area for us. And we've been doing pretty well in terms of expanding with the enterprise clients, with big tech clients. Also from an industry perspective, focused on technology. We are continuing to penetrate into health care and banking industries. Those are some areas that we are focused on right now. From a size perspective, we guided to a midpoint of [ $905 million ]. We have about 47,000 employees worldwide in about 13 countries. And the culture is like a big part of our -- of how -- what TaskUs is, and that's something that we continue to focus on.

Got it. I guess just maybe start off on kind of the near-term demand environment. You've had multiple things past few quarters, some of it client-specific, some of it macro. But just think about last quarter, if you could talk about the factors in terms of what changed and what led to the lower outlook. And so the factors we talked about the quarter before, it was like U.S. delivery, largest clients, something like that. Has that gotten worse at all? Or is it more kind of macro driven?

Right. So maybe -- I'll maybe talk about year-over-year first. So one thing that happened year-over-year is that, we saw the shift from revenues from onshore to offshore. And also, we saw volumes being impacted in our crypto and equity trading clients. So that has an impact year-over-year. In terms of more recent quarter, one thing that we're starting to see is clients being focused from a cost optimization perspective, and efficiency, that's something that companies have been doing for the last few quarters. So we did see some volume impact, as an example, in our top 20 clients. And that's something that we have seen across the industry. It's not something that is specific to TaskUs. So we're seeing that the clients are continuing to focus from an efficiency perspective. But I do believe that we are probably closer to the end of the cycle right now than the beginning of the cycle.

Got it.

I would just add 2 other things really quickly. The sales cycle, we've continued to see a delay in decision-making. And I know, you and I were just talking about this outside. Clients, I think, are hesitant to make decisions right now, given kind of what 2024 budgets may look like. So they're not -- we're not losing deals, but we're seeing these deals get pushed out. So it's one other factor. And the other thing I would say is, clients are definitely embedding more process. This isn't AI. This is more binary process, more legacy tech or more improvement to their actual product. And so I think that's the other piece that I would call out. You're seeing kind of that impact into our forecast as well. The last thing is, just given the volatility we've seen, I think we've appropriately embedded a certain level of conservatism into the outlook. And it's not that we're trying to sandbag or undercut our forecast. I think, again, given what's happening in the macro environment, it's appropriate.

Yes. And just pure forecast perspective, even historically, the in-quarter guidance that we provided, we have always come ahead of what you have said. So we have very strong visibility in terms of the in-quarter guidance. And if you look at full year 2023, half the year is already actuals, and then we have pretty good visibility in Q3. So we feel really confident in terms of the guidance that we provided for 2023.

That makes sense. Double-clicking on the kind of demand pipeline. Is there any differences across verticals, geos or services lines? Like are there certain areas you're seeing greater pressure, some you're seeing maybe some green shoots of demand, or anything like that?

Yes, maybe we'll get started. So we've seen a pretty good growth inside of the technology vertical that grew 50% year-on-year. And like I said before, we have continued to see a lot of momentum inside of health tech. In our earnings call, we spoke about a few examples. So we're seeing a lot of progress there. We're seeing a lot of progress in banking. So those are some verticals. E-commerce and retail, so those are continuing to do well. From a geography perspective, we are growing well in offshore. But if you look at nearshore geographies, especially Latin America, that has grown 70% year-on-year. So we're seeing pretty good growth. And from a service line perspective, I'm sure this question is coming, but we've made a lot of progress and investments inside of generative AI and TaskGPT. Alan, anything to add?

No, I think you had everything.

And you've been making some investments into sales and also like international expansion into Europe and places like that. Have you seen any momentum from that? Or is that still like a longer-term focus?

No. So definitely, like, for example, the investments that we made inside of enterprise sales. And that's one of the reasons how we've been able to build deals in that particular area. So it's already -- we are starting to see momentum there. And we're also making investments inside of specialized service lines and inside certain geographies to make sure that we go after some of these geographies, as an example, Europe. So some are already happening and some, we'll start to see in the next couple of quarters.

Yes. Several of those large global clients that we mentioned on the last call are European-based, actually. So that's kind of starting to show some traction there. The other thing we're seeing traction in is Asia-based operations. So Asia for Asia, right? And so whether that's growing in Malaysia or in Japan or in the Philippines to serve some of those Asian-based geographies for our clients. We've seen traction there as well.

Got it. And I know you're not going to give 2024 outlook here. But just directionally, how are you thinking about the setup for 2024? I know a lot of it is like when sequential growth returns. From a year-over-year perspective, I feel like getting back to closer normal levels, it keeps getting pushed out. But anything like kind of in context when some of the tougher comps from like your largest client and like crypto and equity clients like that when that battle up.

Yes. So what I would say is that, again, we're not giving any guidance for 2024 at this point. We'll do that formally when we do our annual call for 2023. Like I said, 2023 is progressing really well, both from a margin and revenue perspective. But for 2024, what I would say is that, considering that the comparison, especially year-on-year, the comparison continues to be tough because of -- especially even in Q1, we had some impact of shift from onshore to offshore, et cetera. So I'd probably say that we should look at sequential growth rates getting into 2024 because the year-over-year comparison would still continue to be a little bit challenging, especially in Q1.

Yes. The only other thing I'd call out is, while we don't have a ton of seasonality in our business. We do have a little bit of seasonality, especially with some of the ridesharing and food delivery clients, and some of the retailer e-commerce clients. So when you think about growth from Q4 to Q1, just embed that into your thought process as well.

Got it. And also a large investor focus in terms of who's derisked visibility. I guess maybe pulling it back, could you give us some context in terms of how your contracts are structured, what kind of visibility you have one quarter out, 2 quarters out, et cetera?

Yes. So without talking about any particular client, so what we say, in general, if you look at our contract length, they typically range from 1 to 2 years. And then we have auto-renewal provisions, and it's built into the contract. And we also have call up provisions in many of our contracts, which kind of protects us from any wage inflation up to a level. And then typically, you sign a master services agreement with the client. And then you typically sign a particular service line with a client, and then you would have an opportunity to grow with the client into other service lines. And we typically don't see much difference inside of service lines, from a pricing perspective. So the biggest driver of revenues and pricing is the geography in which the work is being delivered.

And how quickly could a client ramp down or ramp up volumes?

Yes. So this is pretty typical for the industry. So what we typically get is probably, on an average, about between 30 to 60 days of a locked forecast from the client. And then, we kind of get a volume outlook, let's say, between 90 to 180 days. So we've constantly communicated with our client base, through our operations team and our client services team, to kind of look at how are they looking at the macroeconomic environment, how are they looking at volumes. So you do get outlook from them, but the lock typically tends to be between 30 to 60 days.

Got it. I know you mentioned some cautiousness in the outlook. But are you doing -- taking what clients are telling you in terms of volumes and then like a risk-adjusted, like a risk-weight adjustment on that? Is it -- or do you have some macro assumptions in the second half of the year? I guess some context in terms of where the cautiousness...

Sure. I'll say it's really both. So we are talking to our clients on a daily, weekly basis. Our client service teams are feeding that back. Our operations teams are feeding back volume information to us, to our forecasting team. We're also looking at the macro environment, the risks that we see for potential volatility or variability in volumes. And then lastly, we also have more of a kind of peanut butter spread, right? You take a certain level of hedge, and you spread it across the entire company. And you say, okay, there's a certain level of risk just given volatility in the environment today. That certain level of risk has increased over time. Obviously in a high-growth environment, you have less of that, and in a slower growth environment or a macro environment where there's kind of ebbs and flows or risk recession or other things like that, you add more risk into your model. And so that's kind of the situation we're in right now.

Got it. On the full year outlook, I guess maybe the low end versus the top end, what needs to happen to hit each side of that?

Yes, I'd probably say, like Alan mentioned, like some of the risks that we are factoring as a top side risk, like let's say some of that may -- doesn't materialize. Then that would probably help us in getting into the top end of the guidance range. Or even the assumptions that we're making from a sales conversion perspective, let's say, we're able to do better, that will be another way to get to the top end of the guidance range.

Got it. I guess maybe shifting to AI, the hot topic. I guess maybe starting on like benefits you guys might see for a minute. Maybe talk about some of your capabilities in AI, particularly AI services, that sort of work you've done there and how the offerings kind of evolved.

Yes. So if we kind of go through maybe from -- like from each service line. So if you kind of look at digital customer experience, the largest service line. So that we believe would be an agent assist tool. So it's not going to replace the agent. While we might see some automation in low-end customer service work, which is something that we don't operate in, we typically do more complex work where you have to leverage other technologies. You're going to be doing multichannel support. So we don't see that getting automated, at least in the near future. So we kind of look at it as an agent assist tool, so which means how do you make the agent more efficient. And that's where TaskGPT comes into the picture, where it's going to make the processes more efficient. And we also would work with our clients to share some of those benefits that we see from a workflow perspective. So that is on digital customer experience. Then if you're going look at trust and safety, we feel like that would be a huge future opportunity where this technology is going to be creating a lot of opportunities in that particular area, as an example, deep fakes, or even the amount of user-generated content that is being generated. So we see that as a future opportunity in trust and safety. And then in AI services, we've done a lot of work in terms of data tagging and data annotation. So we, again, see that a natural opportunity for us in terms of the content that is being generated, especially specialized content. So how do you, kind of, verify that content, create that content? Again, we see that as an exciting opportunity in the future. Anything to add there, Alan?

Yes. I mean I would just -- just to double-click a bit on the digital customer experience, AI, support work and kind of what we're doing there. So -- and this is kind of stepping back when you asked about AI services. We look at generative AI in 2 buckets, right? One is servicing the generative AI companies. That's in the AI service line that we have today. But the other is leveraging generative AI to change the way that we work with our clients. Looking at digital customer experience. And I think, this is where -- and you talk to investors every day, so feel free to correct me if I'm wrong here. But what I hear at least is, the area that the investors see as higher risk is automation around the digital customer experience. I think 2 factors here. First is, Corporate America and corporations across the globe move very slowly. And customer service professionals are likely very hesitant to stick their neck out and say, I'm going to automate everything tomorrow. So I think you have a long runway there or at least some runway there. Like Balaji was saying, there's certain tasks that are very easy to automate through generative AI, right, binary tasks or things that are data-driven. That's not really where we play. But what we do tends to be dynamic, real-time, multivariable, changing variable in real time. That stuff is really difficult for, at least as of today, for generative AI to automate. When we think about the opportunity, augmentation, what Balaji was talking about, is exactly right. We're able to go to our clients and offer them a way for them to service their customers better and save on cost. And the best way for us to do that is to move to an outcome-based pricing model. So -- and it's what you've seen in the BPO space, time and again, as you've seen technology kind of come up through the industry. So as the BPO companies adopt certain technologies, move to outcome-based pricing for those service lines and then are able to capture share because they are moving quickly, they're able to grow. And we are fortunate in that we started working with the world's largest generative AI company over 2 years ago now. So we've been working with this technology for internal purposes for quite some time and have been developing an external offering for a little bit less than that, but for an extended period of time. So again, I think we're one of the first companies to have commercialized it, and already have a client working with -- at least our generative AI offering, and from what I see, kind of across-the-board generative AI offering to augment the customer service experience.

So a follow-up on that, I guess, on the digital CX side. I mean you mentioned it's going to augment agents' capabilities. Looks like your agents might reduce the need for how many agents you need for certain tasks that you're kind of supporting. Does that increase your TAM potential as other verticals or other companies see what you're able to do and how quickly you're able to respond potentially?

Definitely. I think that there's kind of 2 things to think about. One is from an industry perspective, as you're able to embed internal data from a client into the TaskGPT system, which is the generative AI system that we've developed, you're able to become overnight professionals within that space, right? So the training becomes quicker, the reaction time becomes quicker, and we're able to meet the needs of more complex organizations over time, right? So that's one piece. I think the other piece is, again, having the experience. Nobody likes to be the first person to a party, right? And so if you have clients that are already using this technology, and it's proven, you're able to go to some of those new potential clients or new potential industries and say, look, here's the experience that we've had and here's the benefit that our clients have been able to realize as a result of this. So let's walk through what that means for you.

Yes. And also like, Alan spoke about the outcome based pricing model. So one of the advantages there also is that, it kind of makes the relationship a lot more stickier because it's not just not like a human being, but also have a technology claim inside it. So I'd probably say that, with that relationship getting stickier, and once you're able to demonstrate efficiencies and performance, then you have probably an opportunity to increase your share of wallet with that client.

Have you seen increased uptick of the outcome-based models? Or is that kind of more of a longer-term push? Like in the past, clients, some -- we've heard other company say in future we're going to be more outcome-based models, while the clients push back because they also want more benefits of the...

I think when both of us started in the industry in like the mid-2000s, like 2000s, right, like 2005, 2010, people were talking about shift to outcome-based pricing, right? So you're right. I mean the industry has been talking about it forever. We are seeing a little bit of an uptick. We talked about moving one of our larger clients to, at least partially, to an outcome-based pricing model last quarter. We -- our total percent of revenue on outcome-based pricing is in the low double digits. But we do see opportunity. You need to have a certain level of maturity as a client to move to that model. But I think -- I mean I think it's fair to say we're starting to see interest there.

Definitely. I feel like with gen AI and TaskGPT, so that is probably going to be the -- hopefully like the pivot that kind of helps -- help companies kind of move to an outcome-based engagement. And hopefully we see that happening soon.

Got it. And as clients pick up more outcome-based engagements, what are some of the financial ramifications for you in terms of revenue? I assume it's better for margins.

Definitely. So what happens is that, when you typically sign an outcome-based engagement, you would have to guarantee some savings upfront for the client. And in terms of total cost of ownership, how do we optimize that? So you would see some impact to revenues initially. But over a period of time, we would see that there is margin accretion because it's -- because you don't share all the benefits to the client, but you keep some of it. So it's kind of a pretty good alignment of interest between the vendor and the client. And also, like I said before, it also makes the relationships stickier. So hopefully, in the future, any revenues that we lose because of the optimization and the sharing of benefits, we would be able to capture through additional share of wallet.

Got it. I guess circling back to the AI conversation. I know you mentioned like investors focus on digital CX, but also there's been an increasing focus on the content moderation side of the business. And some gen AI companies have -- like [indiscernible] have said that they can use that to help support content moderation. So how are you viewing that context? Do you view it more as an opportunity or more as a threat for content moderation?

Yes. So from a content moderation perspective, to start with, it's always been a combination of technology and human content moderation. It's not been that all the content that is being generated today is being moderated by human beings, right? It's always been a combination of technology and human beings. So there's really nothing new here, frankly. And TaskUs actually has been helping some of these models inside of AI services in terms of training the models. So the second is that while the text-based simple content potentially can be automated, and maybe it's already happening at this point. But if you kind of look at contextual moderation, as an example, let's say, political moderation. You need to have a human element in there in terms of moderation, or even in terms of types of content that is being moderated. As an example, video, audio or even live streaming that is getting moderated. And the amount of content that is getting generated is only increasing. So I feel like it's always going to be a combination of technology and human. It's not going to be -- at least I don't see a situation where it's all going to be completely automated.

And the thing to keep in mind, right now, the technology is really good at text-based review. But like Balaji was saying, when you have 3D images or live streaming or even video with background images, there's a lot that it has more challenges capturing. So that's when you have humans step in. But the other thing is, there's always this idea of escalation volumes, right? So even if an automated system captures a certain image or text or anything else and flags it many times, that needs to be reviewed by a human to figure out what happened or why. So that's the other piece of it.

And to the extent clients want to use more technology in the content moderation side, I'm sure that's a boost for AI services. And you guys launched...

That's exactly right.

TaskVerse platform last year. So I guess -- wonder if you can give an update in terms of how that's kind of evolve the client adoption from the size of, I guess, your client...

Sure. So yes, so the TaskVerse has been -- it's been a great story. We launched our gig worker platform about a year ago. That's right. A little bit less than that. And we've been growing well over 100,000 taskers on that. And as I was actually talking to an investor just yesterday about -- that he had signed up and had an offer for like a $35 thing to review 10 images or something. He was like, "This is great." And he didn't do it. But he was saying it's like the kind of thing that he understands the value and why people are doing this. What I think is really interesting, and this is where I think you're going, is the type of gig workers that we can attract to do whether it's data tagging on images or it's somewhere with a specialized service that we may not need in-house full time but if one of the large language models, the answers get flagged to be incorrect or not ideal around something really esoteric, we can hire a person with 18th century British poetry professional who can do that work and say, okay, I understand why this isn't correct and let me provide the correct answer. So again, it's finding those skilled professionals in kind of one-off situations. That industry is growing tremendously. And we are part of it. Many of our competitors are part of it. There's a couple of private, very large, kind of high-growth companies that are part of it. We do see a lot of opportunity there. But again, the timing of launching the TaskVerse was positive from that perspective. We're well positioned. And again, we're not playing catch-up. We're in a situation where we have a lot of those skills kind of already at our [ deal ].

And especially if you have, let's say, project-based work, where you are to ramp up, and then you have to ramp down once the project is completed, which is typically what we see inside of TaskVerse, it's probably a great model to be able to do that and deliver for out now.

Sure. Yes. Shifting to your largest client. Would you say kind of the shift to offshore we're kind of past that largely? I know that they deprioritized some areas 2 quarters ago. So I guess what types of the work do they deprioritized? And in terms of wallet share of the clients, is there any context you can give? Maybe employee count serving that client, just to prove that [indiscernible] study.

Sure. Yes. So for the largest, the relationship continues to be very strong and we support them in multiple lines of business across service lines. And like I said, Ryan, we did see the shift from onshore to offshore throughout last year, and we also had some impact in Q1 of this year. So the employees that are currently left in the U.S. doing work for their client, but probably a handful of employees, so that risk is materialized and it's already factored into the forecast that we provided. Again, the client -- that particular client continues to be focused from a cost optimization perspective. So they're making sure that they focus on the most important projects and they're deprioritizing projects that are not as critical. So that is again factored into their numbers. From a head count perspective, we did see that, despite the fact that revenues did reduce, we saw growth from a head count perspective because of the shift from onshore to offshore. So we did grow with them from a head count perspective on a year-on-year perspective. And then if you kind of look at the guidance revision that we did this quarter, it was not because of the largest client. It was kind of factors outside of the largest client, which was volume driven for the rest of the top 20 clients.

Yes. And I would just say that, if we look at the pipeline today, there are a number of opportunities with that client that are very interesting, but we're not embedding in our guidance or outlook today. The other thing I would point out is, a big part of the work that we do is political ad moderation. Obviously, we're entering into what will likely be an interesting political cycle over the next 18 months. So we would expect to continue to see increased activity around that.

Got it. Is there an internal focus to kind of decrease concentration, get more diversified in terms of clients and verticals over time? And is there any focus areas or any strategies you're doing to make that happen?

Yes. So definitely, from a client perspective, again, that's something that we are constantly looking at, making sure -- and that's how we kind of do that in terms of looking at new industry verticals. Like guys spoke about health care and health tech and or the -- what we've been doing in retail and e-commerce. So there's been a natural diversification there as we kind of add new service lines. But again, purely from a vertical perspective or an industry perspective, it kind of depends on the type of clients that we go after and we get, right? I don't think there's like a huge focus from that perspective. It kind of we're focused more on the client, and this kind of creates a natural diversification.

Got it. And then I guess going back to the shift to offshore in terms of where you stand with your U.S. delivery. I know I think it's 10% of what you guys assume in the second half of the year in terms of like [ floor gain ], too. So I guess what's baked in those assumptions? Is there additional work that you think could go away to get to that 10%?

Yes. So I'd probably say like it will be somewhere between 10% to 12% in H2 -- rather Q4. And the reason is because, like Alan said, we also have a risk factor that we baked into the forecast, and there is some kind of an allocation that happens. And those risks are not being confirmed by the clients. So we don't know exactly if it materializes, where it's going to materialize. That's why I'm giving a range for the U.S. But what we feel is, again, U.S., too, continues to be an important geography for us and for our clients. And the work that is left in the U.S. is either for regulatory reasons, as an example, health care, which has to be done here. Or there is strong client preference in terms of the process that is being supported out of the U.S., and they want that to be left in the U.S. So I feel like, even in the medium term, U.S. would probably continue to be around 10% of our revenues.

Got it. And network that's left there, do you think it's relatively safe? Or is there still some level of work you think that could eventually move to offshore within that?

No, I feel like it's relatively safe. And also, we continue to win deals in the U.S. It's not like we're not winning any deals in the U.S. So I feel like even if there is some transition that happens, there's going to be some new work that is going to be coming into the U.S. I feel like -- so 10% is probably a fairly stable number that, at least, that we think would be for -- at least in the medium term for the U.S.

Got it. And earlier on, you were talking about enterprise clients got to sign like kind of late last year. In terms of opportunity with them and how they ramp in, could you give an update in terms of where they stand? Have you signed additional enterprise clients and new large set clients in the recent course?

Yes. So the big tech and enterprise clients, again, we made a lot of progress with those clients. Again, the -- if you look at the sales cycle and the actual delivery cycle for the enterprise clients tends to be longer than the rest of these clients in the portfolio. The reason is because, again, they are really advanced from an outsourcing perspective. They have a procurement group, so the typical sales cycle tends to be slightly longer. And then you kind of do like a proof of concept like a pilot, and then you start to ramp with them. I feel like we've done pretty well there. In fact, I'd probably give an example about a large retailer that we started to work with where we ramped up with them for seasonality in Q4. And then we typically see many of their other vendors started to -- started to ramp down post seasonality. They kept actually the work with us. And now we have twice the number of people that are supporting them when compared to last year. So we've done really well and we continue to do well. And again, we're going after some of the other traditional industries. Maybe, Alan, do you want to talk about that?

Yes. So the investments we're making today are really around on the sales and marketing side, targeting health care, BFSI. And this isn't the traditional customer care support work that -- and call center work that those companies invest in. But it's really helping them to take the next step on their digital interaction journey. So whether that's in a chat support or other kind of tech-enabled support work, which tends to come in at a higher price and with a better margin. The other thing that we're seeing traction on, and Balaji mentioned this, and we talked about a bit on our call, in terms of enterprise size clients. So we signed a deal with one of the largest global retailers, actually 2 different ones. One was in the sports apparel play -- in the sports apparel industry, and then one is a luxury retailer. And for both of those, we signed work with a certain subsidiary. But it gives us an entrance into a much larger global conglomerate. And that is the type of kind of foot in the door that can be really interesting for us.

All right. And now you've been investing in -- recently it's been more like nearshore delivery, expanding that. So like could you give an update in terms of, I guess, how much delivery is currently coming from near-shore, what markets you're potentially looking at and how you're kind of thinking about that going forward?

Yes. So like I said, offshore is continuing to do well. Again, nearshore grew at 70% year-on-year. So you kind of split that into, let's say, 2 categories. One is what is supporting U.S., which is nearshore for the U.S. would be the presence that we have in Mexico and Colombia. And in Colombia, we were in Cali, we just opened up an office in Medellin. And then if you can look at Europe, the presence that we have in Greece and Eastern Europe, and some of the capabilities that we got with the acquisition that we did with heloo last year. So that's helping. So those are 2 nearshore geographies or regions that we operate within. And like I said, they're doing really well because for clients that do not want to continue to optimize costs, but also want the employees to be working in the same time zone are able to be able to travel to that location. So I feel like these nearshore geographies are a great opportunity for our clients to work with us.

Is adding additional countries still going to be a focus going forward? Or going back, I guess, to AI, gen AI, does that help augment some of your agents' capabilities to understand multiple languages, for example?

Definitely. In fact, one of the things that we're seeing is that even, as an example, in European support and multiple languages that get supported there, the gen AI's capability of translation is pretty good. So that could be leveraged. And again, as an agent augmentation technology where agents who are not in Europe, let's say are in an offshore location, potentially can leverage the tool and continue to provide support. So that's an area of opportunity for TaskUs.

Yes. And remember, the vast majority of our work is not voice work. So that translation capability through chat is very strong, and you don't have the accent support issues that you -- that some of our competitors may have.

Yes, makes sense. I guess shifting to margins last few minutes we got here. Can you kind of talk through the margin drivers that we saw last quarter and also through the rest of the year, and set up, I guess, 2024 margins?

Yes. So we did -- when we guided margins for -- in Q2. So if you kind of look at H2, the margins are going to be lower when compared to H1. And the reason is a couple of factors. One is the impact from a revenue perspective, which is going down. And second is we spoke about the investments that we're making in sales and go to market. So that's something that we will -- that's impacting margins to drive growth. And the third is that we do see some seasonal expenses typically that we incur in Q4 that's kind of factored in the numbers. So we've guided to about 23% from a full year perspective on an adjusted EBITDA. We did have a G&A optimization exercise that we kicked off last year, again, that's yielding great results. And we'll continue to do that for the rest of this year and also getting into next year.

Yes. I would just say, for 2024, there's a lot of operating leverage on this business. It -- obviously it impacts us in the way down, but it also helps us in the way up. And so as we return to revenue growth, particularly as we look out to the back half of 2024, and we aim to start growing again, margins will benefit from that.

So longer term, not necessarily 2024, there's an opportunity to expand margins from...

Definitely. As I said, we continue to grow our specialized service lines, and we'll continue to grow offshore and nearshore, which like I said before is margin accretive. So definitely there are opportunities if you kind of look at the medium to long term in terms of margin accretion.

And across some service lines, how different are margins between like digital CX versus, like, content moderation?

At this point, the biggest driver of margins and pricing is the geography, where the work is being performed. But having said that, we have started to see that in some of the specialized service offerings that we started to introduce. We started to see differentiation even within service lines inside of the geography. But at this point, the way it is structured, I do feel like geography is the biggest driver. But we -- let's say, go in the future, we'll start seeing more differentiation at a service line level, but not today.

And then outcome-based contracts...

Definitely much higher margins. Yes. Yes.

I guess in terms of offshore versus onshore, what kind of difference do you see in gross margins between the 2? Is it a large difference? I know like overall profit dollars comes down. But just in terms of -- just help us kind of characterize the level of difference there.

Yes. So I don't know if you have publicly spoken about what the difference is. But like you said, the revenues are lower offshore when compared to onshore. But we do generate higher margin percentage in offshore regions when compared to onshore. But we do lose even on margin dollars because of just -- purely because of the impact from a revenue perspective. But the margin percentage is definitely accretive, so which helps from an EBITDA perspective at the company level.

Got it. I guess jumping to capital allocation. How would you kind of rank your priorities there? I know you've done at least 1 acquisition, some 1 or 2?

One.

Sorry. How are you thinking about that going forward?

So if you kind of look at the 3 priorities, I'd probably say investing in the business is #1 priority, right? Like expanding into geographies, as an example, Latin America that I spoke about. The expansion that we're doing in Colombia. So again, that comes #1. We'll continue to invest from a sales and go-to-market perspective. So -- and the second is looking at M&A. But we've seen that the private company valuations are still continue to be elevated and are not close to the public company valuation. So then the third one has been buying back shares, and that's something that we've been executing on for the last few quarters. And we've bought close to about 5 million shares so far. And that's something that we'll continue to optimize and start returning money back to the shareholders. But again, investing in the business continues to be #1 priority.

And share repurchases, do you expect that to be a longer-term focus? Or is it kind of just you taking advantage of the market today?

Yes. So we have a very systematic plan. So maybe Alan just want to explain that...

Sure. So the way we established the plan is a tiered structure where we repurchased more shares as our stock trades at a more attractive multiple on a relative value perspective against the basket of our peers. We relook at that -- those tiers and that multiple every 6 months or so to ensure that it's aligned with where the current trading multiples are. And again, you saw that we increased the number of shares that we repurchased in Q2 when we reported by a fairly significant amount. And that was really, again, as a result of the price declining and us being able to purchase more shares based on that tier structure. It's very formulaic. We're not trying to trade in the stock or anything like that on a regular basis. It's really one sort of a process. And so that's what we've done. I'd say in terms of longer term, ultimately, that's a board-level decision. We had about $125 million left of capacity on our buyback plan at the end of last quarter, and we'll utilize that. The plan goes out until the end of 2024. So we'll utilize that based on, again, that trading strategy, that repurchase strategy that we put in place.

Got it. And not M&A for you guys, but industry-wise, there's been some industry consolidation within the digital CX space. So just any thoughts in terms of what -- how that impacts you guys. Are clients increasingly looking for larger-scale vendors? Or is it more for the type of work that these companies are doing?

Yes. No, I'd probably say like in terms of the conscious effort that we put in terms of expanding our global footprint. And now we have presence in all the key areas, and we are present in about 13 countries. So I feel like we have the presence in terms of going out to some of these enterprise clients, right? But just in terms of the consolidation, typically, what happens with this consolidation is that there's some client disruption, right? Or there could be some disconnects between multiple clients, obviously, being in the same industry and not being able to operate with the same vendor. We haven't seen any major activity at this point, but I'm sure there will be some opportunity that we'll see in the future.

Yes. Makes sense. And we're up on time. So any last thoughts you want to leave people in terms of what should we expect from TaskUs in the next few months and quarters going forward?

No. I'd say, again, there's like a lot of chat about generative AI and TaskGPT. And I feel like, I mean, that will be a net opportunity for us, and we're super excited in terms of the investments that we've made there and the future opportunities that we will see, especially inside of trust and safety and AI services. And the second is just in terms of the impact that we saw from a volume perspective. Like I said before, we do believe that we are probably closer to the end of that cycle in terms of optimization rather than at the beginning of the cycle.

Great. Thanks again, and time for lunch now.

Thanks, Ryan.

Thanks.