Twilio Inc. (TWLO) Earnings Call Transcript & Summary

How is everybody doing, by the way? First day, it's just day 1. We've got 3 more days so energize yourself. It's going to get better and better and better. It's going to be good now, and it's going to get -- not good. I shouldn't say get better. I should be careful about that. These things are webcast, not -- and stuff. It's going to be really -- this one's going to be really, really good because I believe it's one of your first opportunities as CEO to talk to Wall Street at a major conference?

It's one of them, yes.

One of them?

Yes.

But we hope this is the best.

Okay. Great. Well, let's see.

So for those that may not be familiar, tell us about your story, your path to CEO.

It's my calling.

Tell us about it and then we'll jump into the questions that I have in mind. That's my colleague, Matt Martino, the great Matt Martino, we double up. When companies have very complicated products like yours, I need help, so he's going to be assisting me here.

Yes. I mean I'll kind of be brief. I came into the company about 6 years ago as the CFO after a 22-year career at GE, which kind of seems a little bit unusual in some respects. But I'd spent several years out here with GE Digital before kind of the GE story played out the way that it did, but came into the company as CFO, shortly after the company went public, really, as we were starting to scale, I think I was employee 979 or something like that. So there's still less than 1,000 people at that time, less than $1 billion in revenue and then kind of progress through a variety of roles. I became the Chief Operating Officer. And then earlier last year, became the President of our larger business, our communications business and then became CEO in January, and it's certainly my privilege and honor.

How has it been? Did you think you'd be a CEO, one day, of a tech company at all? What does it feel like?

I don't know. I mean, look, it's -- I think you get motivated about delivering technology on behalf of customers, right? And I think Twilio has a really unique value proposition in that we sell really essential and compelling infrastructure, whether it's through communications and data. And did I think I'd be CEO Twilio? Not necessarily. I mean we were kind of founder-led and had been for a long time. But when the opportunity comes along, and I love the company, like I kind of bleed Twilio red, if you will. And it was kind of the right thing at the right time, and I was super excited to be able to do it.

That's great. So I'm sure that you've had a bit of time to let that sink in and put your own stamp and feel of where you want this company to go in the next 2 years, 3 years, 5 years. Where are you in formulating your audacious goals for the company as CEO?

Well, I think, Kash, I mean, I think it starts honestly, like, with the last 6 months, right? So let me start short term maybe before I go more medium term. I think as Aidan and I kind of looked at the company when I came on, it was like, how do we just fundamentally run this company better. We have great assets, but I think sort of shame on us, like historically the company hadn't been run as well as we had the capacity to do. And so we spent a lot of time, especially, in the early days, like really installing what we've now been characterizing in the last few calls with investors as financial discipline, operating rigor and then innovation focus. And, like, what does that mean other than sort of the headline words? Like, in terms of financial discipline, we took a company that just a couple of years ago was losing money, consuming cash and is now mid-teens operating margins, non-GAAP, and with a marker out on GAAP profitability for next year, have reduced our stock-based compensation considerably and have a path to continue doing so, generating a lot of cash and we feel really good about that. The way that we're running the company day to day, I think, is much simpler, much more methodical, much cleaner kind of internal operating rhythms, trying to be much more transparent with our employee base as well as our investors and with customers at the same time, by the way. And then finally, as it relates to innovation, the way that I kind of characterize it is, is that if before we were focused on 10 to 15 things all at the same time, all with equal intensity, today, we're focused on 5. We're putting the full kind of muscle of the company behind those 5 with the expectation that a handful of those things, like, really lead to reacceleration of growth. So now back to your question over the next 5 years, I mean, I think those investment bets that we're placing, like, they're not super far afield of what we do today. I think they rely on the assets that we've got, a leading communications company, leading data company, and, I think, as we look forward, the combination of communications plus contextual data, plus AI is a winning hand. Because, like, both of you, you can kind of think about it through a lens of your own life. Like, the number of amazing digital interactions that you've had with a brand over the last, I don't know, 2, 3 years, like, you can probably count them on one hand because it doesn't happen that often. And I think it's not just about personalization, but it's also about like real demonstrable ROI that we can deliver in cost savings to one of our customers that they can see and measure and/or revenue uplift that they can see because we're creating a much more intimate experience between one of our customers and the consumers that they serve.

Got it. This is super interesting. And so I know -- [indiscernible] it, but since you said you focused on fewer outcomes, right?

Yes.

Fewer bets, what are those bets? I mean, do you look at it as product X, Y, Z or initiative XYZ? What are the more concentrated things you're doing as opposed to a lot of diffused things that might have been pursued?

Yes, I'm not going to name them, but, like, we'll do an Investor Day at some point next year and share some of those things. But I think fundamentally...

See, I asked the right question then. See?

I mean fundamentally, they're geared at -- think about it this way, okay? The LLMs are -- they're amazing, right?

Yes.

And they're super powerful. We're not in the LLM business. But what our customers care about and what they will never give up to the LLMs is proprietary data, like the contextual data that they have about their consumers, they would be bananas to be able to turn that data over to an LLM to train it up, right? So that's not going to happen. Instead, they want to use that data inside of their data warehouse providers. They want one copy of it, right, because their CISO says they don't want multiple copies floating around. And to be able to transact appropriately with the consumer on the other side, you've got to have communications infrastructure. Now this has sort of like long been the promise of Twilio. And I think over the last 6 months, like we've actually started delivering products that deliver on that promise. And I think as you see us build out and innovate over the next kind of 12, 24 months, that's kind of the strike zone you should expect.

Yes. So that's great. You laid the path, right? So -- and you said they need a lot of Twilio to do that. Tell us how. What would they be using from Twilio to accomplish their generative AI, I think, let's say, to train their models inside the firewall...

Well, think about it this way, okay? So there's, like, I don't know, several thousand vertical AI companies that are being built, like, right now, right, or watches as we speak. And they will serve a vertical niche. They'll use an LLM and -- but to really activate with their consumer base, no matter what, they've got to ride on infrastructure, okay? That's where communications come into play, right? And many of these companies, my personal view is, is that you're going to see a real renaissance in voice interactions because conversations like this are normal. Like, if you and I were texting back-and-forth, it'd be kind of weird. But it's super normal in a voice environment. The AI is very good at, like, kind of picking up the signals, reducing latency and using contextual data and our voice APIs, let's say, and then their LLM, they'll be able to drive a superior customer experience on the other side. Now what would -- the interesting part about that is, is that we'll be the platform provider underneath and then they'll be the last mile of value onto the consumer. No one ever has to know it's us as long as they do. And they're delivering superior value, but it has to be enabled by communications and AI. So that's, like, one example in sort of a very recent AI context. Now contrast that with, like, I don't know, like a kind of a more traditional industry like retail, right? Like retailers, relatively thin margins. They're looking for all sorts of different productivity opportunities. And so here again, if they can use AI as a means to, let's say, suit their contact center needs, that reduces their cost base significantly. They can plow that investment back into other areas. They can have the human intelligence serve expert services but again, no matter what, in any of that, you have to have communications capabilities, i.e., Twilio. You have to have contextual data capabilities, i.e., Twilio. And then whatever LLM they want to pull in, we're ready and willing to work with anybody.

Yes. Matt's giving me the look, like, "Dude, let's get onto the real questions.

No, no, no. I think you guys have been talking a lot over the last couple of quarters of kind of marrying kind of the Segment business and the CDP business with kind of the core communications business. I think you're kind of referencing them -- referencing that with the contextual data and the communications piece. So I guess how far along are we in kind of integrating these 2 businesses? What can we expect in the future? How material can this be to your kind of future growth prospects?

Yes. I think we're getting there. So in Q1, we launched an Agent Copilot with a unified profile, right? So in the customer example that we gave in that environment, like what happened is, is that they were running on sort of a traditional contact center. They deployed the Agent Copilot. They were running at extraordinarily high CSAT levels. I think they were, like, 96% or something like that. They dropped 1 point in the first month, which is, like, barely discernible and went right back up to 96% in the following month. Okay. So no difference as a result of using Agent Copilot, enormous productivity as a result in that environment. Combined communications, our Flex contact center and the CDP, which kind of underlies that experience, what's cool about it is, is that the consumers that they're serving, all of that data is being harvested in the unified profile. Next time they call, that data is being reused. It's accelerating workflows, and it's getting them to their value points a lot faster, okay. And then in Q2, we launched, within an IVR context, something similar, where again, we wanted to layer in the unified profile as a means to deliver the IVR capabilities, and we're actually now piloting that with food delivery company who wants to be able to use that to kind of, a, accelerate their volumes; but more importantly, actually scale all of the interruptions that happen inside the store as well as upsell customers into something new. Interesting about that is, is that our voice intelligence product harvests all the keywords. The transaction is driven by AI workloads on our voice API. And then all of the interactions that take place there allow for that customer to upsell the consumer into ultra-high-margin products, which are super hard to do through a person because they're trying to rifle through an interaction. And on the other side of that, we can send -- and are, in this pilot, sending the customer an e-mail thanking them for their business, pushing out a new coupon, referencing all their history, creating that, like, really unbreakable bond that's kind of long been promised between the technology that's deployed and the consumer. So those are the first 2 entry points. And I think now the goal is rather than introduce kind of like these unique products, develop the platform full stop, so that any communications, workload ultimately consumes data, informs it to drive personalization. Makes sense?

Yes.

Very interesting. And I want to maybe kind of bridge that with maybe kind of you guys pulling back from this Twilio Engage dynamic. Like, can you speak to, a little bit about, why you guys are stepping away from that? Because it does sound like there's a little bit of overlap in terms of what you guys are trying to accomplish with kind of your 2 product suites.

Yes. So what we pulled back from was what we referred to as Engage Premier, okay? And it was really meant to be the marketing technology that's delivered on the other side with a lot of other like sort of MarTech companies end up doing. And from -- we pulled back, really, for 2 reasons. One is, is that Twilio is at its best when we operate as a platform company, okay? We sell communications, infrastructure and APIs. We sell data infrastructure and APIs. We are very, very good at abstracting complexity from that level. And there are other layers in the infrastructure and platform stack that you can participate in without having to be an application company. I would say the other side of that is when we're an application company, we did not perform very well. And so as Aidan and I kind of looked at the portfolio, like, it was important to, like, stop doing things that you're not very good at, number one. Number two, the kind of tougher thing about that particular product was it put us at odds with some of our largest customers. Like, that's a terrible setup. Like, why go to war with some of your most important customers in businesses that they're super successful in and have created by writing on your rails in the first place. That didn't make any sense to us, right. And so we've kind of taken what was a negative and flipped it to a positive. And in the last earnings call, and as I'm sure you guys are aware, like, we've actually, like, furthered our partnerships with 5 that we named as well as others within the ISV community where we can grow our businesses together. They can use our infrastructure, we can co-sell their products. A lot of these guys are actually not in as heavy as competition as people sometimes think. And we have an opportunity to grow our businesses together as a result.

Maybe just kind of sticking on the communications side. And you spent, a lot of time, in this business specifically, right?

Yes.

And I think investors are always trying to get a sense of kind of how penetrated we are in terms of kind of the SMS messaging opportunity. I think we can appreciate that there's a lot of macro crosswinds out there that may be impacting the usage revenues of the communications business. But more broadly, how do you think about the TAM of messaging? Because I feel like the value of kind of being on that lock screen is really critical. So talk to us a little bit about that.

So I agree with the premise that you stated there at the end. I think the key is this, right, we were talking to an investor earlier in the day, like, who -- and I think it was in India, which is not a market that we participate in, but was receiving 17 messages, right? That's pretty annoying, actually. You don't want to receive 17 messages all from the same provider. So I think the key from our standpoint is, is that for the channels to remain vital, the information that you receive has got to be high quality. Okay. So quality is sort of through the customer's lens, they want it to be delivered and they want it to be delivered at the exact right time that it needs to be. That's why we charge a price premium because our technology is better at that than the competitive [ scape ]. The other side of that, though, is, is that this is where kind of Segment comes into play. How do you make it fundamentally more personalized so that those interactions feel special, or at a bare minimum, is something that you feel compelled to respond to. So now back to your question, I think whether it's Gartner or IDC or Forrester, like, basically everybody is projecting that this -- I mean, this TAM has been growing for 10 years, right? And I think everyone projects that, that TAM continues to grow. I have that optimism as well. I think, actually, what happens is that the number of messages actually probably continues to go up, but they just get a lot better. And the stuff that you're getting spammed on, some of which include some political messaging day-to-day, like, that is going to get annoying and that's the stuff that will start to fall away.

Understood. And then maybe on the Segment piece, right? You guys have been managing through this kind of period of contraction. So what is it going to take to kind of return that business to growth at this juncture?

Yes. Good question. So I think there -- we have seen some green shoots recently. So in the last call, we kind of referred to some interesting stuff that we've done in terms of accelerating time-to-value. So go back a year, it would take a customer to get to value, like, 9 to 12 months. That's a long time. And I think in the most recent examples that we've had with customers, we've got that, like, sustainably down to, like, 30 to 90 days. Like, that's pretty exciting, right? If you are our champion and you're in the environment, you got to defend the purchase to your boss, like, you're going to get the value within a quarter, basically. So we've been able to do that. Paired with that, we've also been able to kind of adjust the nature of the contract, sounds kind of mundane, from being single year to multiyear. Now marry those 2 things together. You're in a 9- to 12-month time-to-value, you're on a 1-year contract, you're not super excited about, like, renewing with segment on the other side of that. So those 2 things together, I think, give us a lot more ballast in terms of the way that we think about, like, sustainable DB&E. And then, I think, on top of that, we put a lot of work into, like, customer success to making sure that customers kind of feel loved through the deploy. And I think we've got, like, a lot of recent examples. We've got to burn off, like, some of the older ones, but I think it takes a bit of time for us to kind of get back to, like, reported growth. But I think those green shoots, they are starting to show up in the business. I think that we'll start to see it initially in new bookings and then it's just going to take time for it to translate to growth.

And how do you balance that hiring or incremental capacity from a sales perspective relative to this breakeven target you have out there on Segment?

So we're very much on track for the breakeven target that we definitely view that as like a red line investor commitment, and we're going to make that. We just don't need to add a lot of head count right now. Like, we -- in fact, we've contracted a little bit on our head count, if anything. We don't intend to add head count. I think -- as a matter of fact, like, we think that there are additional efficiencies that we can go get in the business, whether it's geodiversity, in terms of, like, moving from, like, high-cost regions to lower-cost regions. I think some of these AI workloads that I talked about in terms of, like, us being able to field the solution and drive customer productivity, like, you can be certain that we're driving those same workloads internally to drive productivity. And then, I think, good old-fashioned automation. Like, Aidan starting some things in accounting and IT, stuff like that. And I think all of that means that we can continue driving margin expansion even in a more dynamic macro environment. Of course, we aspire to be double digit.

You called out your CFO 3 times, so Aidan, good job. That's the CFO, Aidan, here with us. You talked about deploying Gen AI internally. Tell us what you've uncovered. What is the unlock that makes you feel optimistic, that is, if you feel optimistic about generative AI?

Yes. I'm very bullish on it. I think I'll give you maybe 2, 3 examples. So this is like, I don't know, like, 1.5 years ago now, something like that. You may remember, like, we were going through this transition where we had to register customers onto 10-digit long codes, right, and pretty painful process like pretty laborious workflow...

My son tried to do that, like...

So what we did was we had an effort that we were undertaking where we had about 100 people working on it over the course of 3 months. okay? It was kind of slow going. And we were sitting in a meeting, kind of monitoring progress and I asked our engineering team. I was, like, surely, you guys have experimented with some AI or ML, like, to be able to maybe deploy alternatively in this process. And they were like, well, it's kind of on the shelf. Like, it's got a high false positive rate, et cetera. And I was like, okay, but it's either that or, like, we don't make it on this deadline that we put out there for ourselves. And so I said, okay, dust it off. And they came back in a few days and they said, okay, we think we can run it. That 3 months of work, 100 people, we recreated in about 10 seconds. Like, literally 10 seconds.

What was it again?

It was a machine learning model that basically took the workflow that you had to go -- there are 2 workflows that you had to go through to get through a 10DLC registration because it had top partners and it pre-populated the entire thing based on some segment data that we had about some of these customers and then got them to the next workflow based on what we would basically serve up to another provider. They're definitely a pass, put a lower bar on them. And actually, we had way fewer false positives than we thought and the lower bar turned out to be a much higher bar than the other provider was using in the first place. So that's one. Two is, as you can imagine, in customer support, we're deploying it -- we have an agent -- a virtual agent that we've deployed, which handles inbound leads, very effective, obviously, scales infinitely super cheap, is able to route to experts if a customer wants to scale up or if it's a particularly complex question. That's been pretty interesting. And then inside of our product suite, something that we've been doing even before you get to -- although this actually is a segment, too, but about a year or so ago, you probably remember this, Kash, like, we launched a product called Fraud Guard and Verify. That all uses AI, where we're picking up fraud signals in the ecosystem. We put a 100% guarantee behind that product. To my knowledge, we've never paid out on it once because it's very, very good at rooting out fraud for customers. And actually, it's interesting because what happens is that with customers, we send them an e-mail every month, showing them exactly how much money they saved and the ROI on that product is, like, 30 days. So it's really compelling.

So you ought to come back -- and we hope that you do come back next year for this conference. We're going to be talking more about how Gen AI has become more mainstream internally to Twilio, right, and externally to your customer base.

I hope so.

I thought you're going to say it, absolutely, yes.

Well, I mean, look, I think the internal proof points are already there and the external proof points, I've alluded to several of them a little bit anonymized, but I mean these customer examples are real. I think the interesting thing about Twilio and the way that we're positioned today is that in driving these workloads, the infrastructure must encompass communications and contextual data. We have both. There are just not that many companies that possess both of those assets. And the LLM stuff, if we don't need to build our own LLMs. Like, we will let -- and we want our customers to use whatever LLM they're comfortable with, and their ability to kind of use the LLMs to drive the workload, but the contextual data to actually drive this personalized interaction, like, that's where it gets interesting. And I think we're already seeing it, like -- so I think the only thing that's going to happen is we're going to see a lot more of it. And I think we'll be a definitive AI beneficiary.

The internal stuff, Khozema is -- are you -- did you build this yourself? Are you using third-party?

We largely built it ourselves. In some instances, we have used partners with some internal stuff. We've also used partners with some external stuff, but a lot of it is kind of home we're in. I mean, honestly, a lot of it, Kash, is like engineers experimenting and it's not quite ready to productize, that sure is ready for internal use, right? And we just sat through like a series of demos. We have this week that we kind of dedicated to this stuff last week. And the stuff that these guys are uncovering is -- it's unbelievable.

Aidan must be excited. Like, yes, look at the basis points of margin, like...

Well, I think...

Is that -- is it too early to have an impact on the operating efficiency...

I don't think so. I think over the next couple of years, you'll definitively see it in our P&L. No question.

Wow. Yes. 200 basis points? 300 basis points? I'm talking like a...

That, we are not committing to today but...

Yes, I'm kidding...

You'll definitely see, like, material margin expansion in our P&L.

Yes. This -- is just in the course of today, I mean, you're the -- I interviewed 8 CEOs today, all CEOs, ninth, everybody is saying the same thing, that internally, they're seeing a lot of success. Started with MongoDB, [ Dave ]. Then we had Ash, we had [ Vlad ] -- everybody is saying the same thing. So you guys are slowly unlocking things that our investors probably...

I would -- the only thing I would edit in what you just said, I think we're quickly unlocking it, actually. I mean, I think these things are happening -- well, I think these things are happening very fast internally. And I think they're really accelerating in terms of the external environment as well. I think the customer -- there's like a customer receptivity thing, right? Like, the regulated industry is a little bit slower, retail, e-commerce, travel and hospitality, moving super-fast, right? And the margin structures of these businesses, in some respects, dictated regulatory structure of those, too.

Got it. And I have a final one for you. Does anybody have a question? Want to be a prompt engineer and ask a really, really good question? It doesn't need to be a really, really good question. It can be a good question, not to raise the bar. 1.5 years ago, you made a big shift, had a big buyback. Tell us more about the capital allocation philosophy. Why do you think the stock's undervalued. You've got a lot of money on the balance sheet.

Yes. I mean, look, the reason we're buying back so much stock is because we believe the company is undervalued, right, I think full stop. I think we still have a lot of cash on our balance sheet. We've done a material buyback program. We -- I think we're at about 2 or so of stock buybacks that we've done kind of cumulatively to date. We'll finish the authorization that we currently have set up. I think heading into next year, like, we haven't made definitive, just, policy decisions about our capital allocation philosophy. But I think what it does mean is that we have a lot of optionality. With the amount of cash that we're generating, there's a lot of different things that we could do. The only thing I would rule out is we're not going to look at transformative M&A. That's not on deck for us. I think we'd be open to looking at like tuck-ins and stuff like that. And obviously, we have capacity to buy back more stock.

That's great. Matt says, I want the last question. I'm going to make it a damn good one.

It's an easy one. No. Look, you guys have done a great job on the margin expansion front over the last 2 years. I know you have Segment breakeven target out there. You've got the GAAP profitability target. But if we look out over kind of a longer arc, I mean, kind of where are the efficiencies that you guys can realize to continue to expand your kind of margin profile on a longer-term basis?

Yes. So I think in the kind of short to medium term operating margin side, I think, again, geographic diversification, AI internally, automation internally, like, we see genuine market expansion opportunities there that, I think, are going to be really material for investors. I think on the kind of gross margin line, like, especially as we start to bring these products together, deliver superior ROI for our customers, I think, increasingly -- it'll take time, okay, but margin rates should start to creep up. I mean, the thing with Twilio is like the weight of the business is, in this communications, half oriented towards sort of SMS, and so that's going to drag down the margin rate structurally. But I think over time, these higher-margin projects, their products will start to inflect that rate and help us over the long term.

Fantastic. We're on to finishing our final fireside of the day. The 10th unfortune to CEO to listen to my questions and Matt's questions. But listen, this has been...

Great.

A lot of fun. Thank you so much. You tell a good story, and we wish you really well, and you got a great CFO to aid you in this journey. Thank you for not asking questions. You should have asked questions, sorry. At the end of the day, I get it. But we hope to have you back next year.

Great. Thanks. Okay. I appreciate it.