Twilio Inc. (TWLO) Earnings Call Transcript & Summary

First of all, how amazing is this conference? How've you guys been? Day 1, day 2, day 3 is the best. Day 3 is the best, right? Actually, it is absolutely the best.

The energy is really high.

Well, that doesn't say anything, because we have a fairly quiet audience even when this room is packed. There's no questions, et cetera. So -- but the quality of the companies, the content has been just off the charts. Jeff, thank you so much for making it to the conference. Really appreciate it.

You were here last year, and I asked this question. What is the long-term vision of the company? And you haven't really had a very compelling answer. I wonder if that answer has changed. So what is your -- what are your goals for the company? What is the strategy? How do you want this company to look like 5 years from now?

So the goal of the company is unchanged, right? We see the ability for technology to completely change the way companies go about building their relationships with their customers. And now I'd say, if anything, that vision for what we can accomplish has been supercharged by all the progress we've seen in generative AI. So at our conference just 2 weeks ago, we really unveiled our vision for how we think gen AI plus the CDP, so taking all the understanding a company has about its customer as told by all the data pulled into profiles of customers and then audiences of similar customers. Gen AI can take that and turn it into more powerful, more effective customer engagement and customer communications through every step of the customer journey. And the end result of combining customer data communications and AI, we believe, is going to be companies will be able to build 10x better customer relationships at 1/10 of the cost of today. When you think about these use cases, right, how would that work? Well, is gen AI able to take a really like all the data about the customer base that a company has and unlock a 10x more performance marketer? Absolutely. Easily. And we laid out that vision. Like AI can actually automatically create marketing journeys for the marketer. In fact, we launched a product called Generative Journeys. You described to the AI, you say, create a journey for customers who haven't bought anything in 30 days and have things in their shopping carts and -- but have bought something at full price at some point in the last year, and the favorite color is purple, whatever, and then try to sell them what -- and you just describe it like that in English. And it will draw the whole journey, like all the boxes and all the, do this, wait 3 days, do that if they did this, do that. It will automatically select the customers that are the best candidates for that campaign, and you just click approve and it starts. And it writes emails too. I mean this is all possible. And then once you say like, okay, that's possible. This is literally something we announced 2 weeks ago. Once -- it's that possible. Once you go from, oh, it takes a minute to do this journey that before would have been probably weeks going around to, okay, we've got to get the data people to give me the feed of customers, then I got to go to the creative writers, the copywriters and then the design people and all that. Now you can just do it in like a minute. Then you start to ask questions of like, okay, well, what's the limiter on how many of those you can have run, right? Because it used to be people's time was the limiter on how many campaigns you can be running at one time. And now you're like, oh -- by the way, we also announced a product called generative audiences. So -- or predictive audiences, where it will basically say -- you wake up in the morning and it says, hey, we analyzed -- AI had analyzed your -- all your customer data and realized that there's a propensity for these customers who bought this product and this and this to go buy that product. They're 40% more likely to buy that product. Do you want me to create a journey on that? Yes. Okay, Generative Journey created. Like so it's something like, okay, AI can have the ideas for the campaigns. AI can actually execute the campaigns. AI can see the full end-to-end of like, is this campaign generating sales? And if you feed it gross profit data, is it generating profit? And if yes, great. Do more of it. If no, slow down or stop it. Now you're like the limiter is if a company typically would have maybe 100 of these campaigns running at a time because human beings had to go do this work, in the future, companies going to have millions of these campaigns running simultaneously. For millions of simultaneous campaigns that are constantly coming up and being created and killed and created and killed, and AI measuring their effectiveness and the human beings are going to look at charts and they're going to test hypotheses. If I feed the AI this data source, does it get better or worse? And so it becomes data science back to the marketers, right? So when you ask the question, can the marketing team get 10x better? Minimal. Like 100x better is probably more likely. And can they do that at 1/10 of the cost before? Easily. Right? So that's the vision we've put out there. And the same thing, by the way, goes for service and support and presales and a lot of other functions at the company, where you're just going to get these 10x outcomes at 1/10 of the cost. That's the power of combining the data that a company has about its customer with generative and predictive AI and then all of the channels to go and deliver on that promise to the end customer.

Got it. Two things that come to mind, Jeff. One is doesn't this lower the barrier to be in AI-driven applications? Because you have data, Salesforce has data, ZoomInfo has data, Adobe has data. All of a sudden, we've gone from the stovepipes of very domain-specific -- you know, I do creative really, really well. I do sales and you do CDP and Flex really, really well too. Oh, okay. So AI is helping us get into the data and draw these deep insights. So you deliver your spiel, you like a marketing automation company, right? So pros and cons. Does it not lower the barrier for anybody to get in and do this? And what is so unique about the data that Twilio has that...

If you think about it, like the core of that story is data and the channels. Like everything else is like largely automation and a few dashboards. And so what it really comes down to is if every company has access to the same language models, and the language model space is commoditizing rapidly, so I think everyone will have access to language models. That's a no-brainer. And so what it comes down to is what proprietary dataset does a company bring to bear that is better than, say, their competition. And so there'll be a variety of answers to this question. People have proprietary datasets, it's about various things on product or maybe I've got data about some domain-specific thing. But every company has customers and every company has customer data. So the question becomes how well do you use that customer data to turn it into these AI outcomes? And there isn't the thing. There's no one source of data, right? Because you take all those, I think you called them stovepipes, I haven't heard about -- like the silos of data that exists among web data, mobile data, e-commerce data, marketing data, purchase data, returns data , I mean, everything you can imagine, but you have to bring it all together in order to unlock this outcome. Just having a bunch of separate data streams sitting in different places does not allow humans or AI to have an insight about what would you do for Kash? What is Kash interested in? What's he all about? You need to take all that data together in order to draw the inferences and make predictions about each person and then cohorts of customers and things like that. That's what the CDP is needed for. And so what the CDP allows companies to do is to tear down all those data silos that are essentially the blocker to getting these AI outcomes. And that's why Segment as the leading CDP in the market, both in terms of its capabilities, but also in terms of its market share, according to IDC, that gives us the edge of helping customers to create their proprietary data asset here that allows them to unlock this outcome better than their customers. And you think about the incentive for companies to do this well, right? If you can get 10x better outcomes at 1/10 of the cost, which I think is directionally what you can get at a minimum, then there'll be so much incentive for customers to get this right.

Got it. That puts you in a new category, in marketing automation and things that you [indiscernible].

It's actually not about marketing automation. It's -- like I actually think that all of this comes down to data, basically. But yes, we are in the marketing automation space with our Engage product, which is built on top of Segment. We are in the contact center space with Flex, our contact center product. But those are all built around the core of communications and data, which is really the foundational assets that are most important in the story, I believe.

Got it. Before I turn it over to Matt, one other thing is -- when you look at the synergies -- in this AI world, what are the synergies between the communications business and the applications business? And what are the things that you can -- you talked about a data-centric view of building these new generative AI applications. What else can you leverage in the communications platform?

Well, you know what, every customer that comes to us for communications is coming to us for communications for a use case, right? Think about the text messages you might receive from a company or the phone calls or the e-mails you might get from a company, right? As the leading platform for voice calls or voice messaging and e-mail, we see customers come to us for contact center use cases, for sales and service, for marketing use cases, for security, like one-time password use cases and for notification use cases. Like those are the 4 most prevalent use cases that most companies come to us for. And so if you think about it, what we've been doing is aligning our stack to enable us to go deeper or really vertical in those areas. For all those contact center use cases that customers come to us for, well, obviously, we have Flex. For all those marketing use cases they come to us for, we have Engage. For all the one-time password and notification use, like we've got our Verify product. So if you think about it, the way we built out our product portfolio is to say -- if a customer comes in and says, hey, look, I'm trying to plug Twilio into this use case around verifications, then we say, okay, well, you can do it yourself, and here's other APIs, or we've already have that built for you that you can plug in and you can choose. And so for us, the opportunity is to upsell customers into a solution that solves the problem they're going after versus building it themselves but still in a very technical way and still in a very API-driven way. And that's the basic interconnect between the raw communication capabilities that we have brought to market that is a $3.5 billion business for us and we're the leader in, into the next areas that we're taking the company.

I have a macro question for you, and then I'll turn it over to Matt. Our Chief Economist, Jan Hatzius, he's been calling for a soft landing, and he's been right so far. We've reached a point where inflation has come down just about enough and rates seem to be stabilizing. And we've gone through a tumultuous 18-month period, and nobody anticipated that we'd have such a massive rate increase. And the economy has hung in there and done okay. Everybody's fears of a recession have proven to be sadly, maybe for them, untrue. The industry has come out okay. So it's been...

Only thing said at banking conferences. Sadly.

Yes, I know. Yes, sadly, for the people that made those negative [indiscernible]. Happily for us, we have a more sanguine view. Where I'm going with this is if things were to stabilize, and we do have a soft landing, we actually call it -- we stole Jan's soft landing, we're calling it software landing. That's a Goldman Sachs proprietary thing, software landing. I'd say if you have a software landing and rates modestly come down second half of next year, whatnot, what does calendar '24 look like in the event that we don't have this massive instability, we just have a more normal economy where people are less concerned about -- what do budgets look like? What does...

Yes. I mean I don't have a crystal ball. But what I would say is you've seen over the last 18 months a lot of companies looking at their investments with more scrutiny and asking what's the ROI of the various software products that we've invested in, trying to harmonize those investments and say, hey, do we have 2 or 3 of this category? Maybe we can consolidate down to 1, or saying, hey, do we have -- someone bought this thing. Are they even still here? And are we getting the value out of it? So you just have those things going on. And by the way, those are good things to go, like a company should be in terms of efficiency. And I think that's what this side of the cycle is actually about. A lot of the times, it's just kind of clean up from the good times of the last cycle and then the whole thing repeats again. So for software companies, though, I look at what -- most of the software companies that I think of. I think about the ones that we use, that we're buyers of. I think about the ones of like other folks I know running software companies. I mean software fundamentally makes companies more efficient. Like that's the role that software plays. Now you actually have to implement them correctly, right? You have to get use out of them. But ultimately, software has the ability to make companies better. Like I believe that. And so I think what you get back to is more of the upside thinking of, hey, if we have this piece of software, is it going to help us drive more market penetration, more competitive wins and put us in a better stance for a longer period of time? Because right now, companies are in that retrenching mode. They're thinking about short term, short-term outcomes. How do I save some money now, right? Whereas you start to get back to thinking about long term, oh, if I invest in this, I will see an ROI. And so to me, that's what the software landing probably looks like is conversations with customers start to turn back to, hey, what's the opportunity to build a really strong competitive stance for the company, because you're investing in software? And now, especially because you're investing in AI, how is this going to evolve the competitive dynamics in your market customer and how can we help you to go win in that world? And I think that's what normalcy will look like again.

So Jeff, just pivoting back to generative AI for a moment. We talked about the implications of generative AI on CDP on marketing campaigns, et cetera. But if we bring that to kind of the Flex contact center product, right, contact center seems like an area that's ripe for disruption with generative AI. So can you talk about some of the announcements coming out of SIGNAL, whether it's Flex Unify or Agent Assist, and maybe how that may catalyze broader adoption of that product?

Yes. So one of the things that we are preparing for is for the contact center to pretty aggressively turn into an automation platform. Today, contact centers are largely human workflow products, right? So human beings log in and take calls and have to browse all sorts of myriad internal systems to help customers, and I think the contact center of the future will be much more about managing the automation than it is around actually having human beings have those conversations. We were talking to some industry analysts recently, and I asked the question, I said how many seats are there globally for the contact center? And the answer was directionally about 20 million. And I said, how many folks -- and these are the experts. These are the industry analysts. How many folks here believe that, that number is going to be more 5 years from now? How many hands do you think went up? Zero. Okay? How many of you think it's going to be less? All the hands go up, right? So then the question becomes, okay, so we all agree how much -- the question is now, how much like -- half as much? 1/4 as much? I won't spoil what the hands did in that part of the conversation, but this is an area where I actually think the fact that Flex is a more nascent business for us is actually a benefit. Because if Flex were a substantial part of our revenue today, we'd probably have a lot of conflict. And I think a lot of SaaS companies have a lot of company -- will have a lot of conflict around AI. Because if you're a SaaS company that's monetizing per seat, you're not going to tell your customers, hey, I think we're going to make your -- people that buy our seats, like, we're going to make them 10% more productive and the future is going to be glorious. Where in fact, the reality is a lot of AI will make it so you need fewer to those people. Maybe 10% fewer, maybe 50% fewer, right? That's the context. And that's every SaaS category, basically, with seat licensed. And so the question is, if companies have built their business model on seats, they're going to get disrupted because they will be hesitant to deploy that technology that requires you to have half as many seats. But that's what the technology will enable. And so disruptors are going to be able to come in and say, guess what, I can offer you technology that means you need half as many seats in a way that the incumbents will be hesitant to do. And so I think that will be true in the contact center as well. And so we are very focused on automation. We are very focused on bringing together the customer data together with those communications channels because, again, I think this largely becomes much more of a data and a data science problem versus A, how do I manage 1,000 people sitting in cubicles problem, and then bring the data together with, of course, knowledge about the company and its policies and all that, those are the foundations of that automation. And you can see we launched this vision 2 weeks ago with CustomerAI, which basically said, look, if you're trying to build AI that is going to serve customers, whether it's in a presales context, whether it is in a support context, whether it's in any of these contexts, the key ingredients here are going to be one, an LLM that has the ability to reason. And that is the amazing thing about the language models that we see today. Like that's the thing that changed in the last, give or take, year, is you now have computers that can reason their way through hard problems. That's why I think this is not hype. This is truly going to be an innovation that is roughly the scale of the Internet in terms of impact on humanity. Because for the first time ever, we've built a machine that can reason. And when you look at some of the problems people have asked LLMs to solve and you watch it work its way through how to solve this problem, you're like, that is reasoning. We don't know how it works, but we also don't know how these work either. So it's the same mystery to us, but nonetheless, the outcome is pretty darn similar.

Yes, so Jeff, I think the common thread in this discussion has been just the massive productivity unlock across all these different categories that Twilio plays in. So I guess, how do you think about pricing for this productivity improvement?

So I think what we're going to do is like clearly move away from per seat pricing. So the only product that we have that does have the option of per seat pricing is Flex. And I think that's just a pricing model that for us and for anyone who's smart will go away, and you'll price it more on probably on a conversation basis. Like think about Segment. The way we price it today is on monthly tracked users. So it scales to how big a customer -- is your customer base. That's, roughly speaking, how much work we're going to do on your behalf. It's also roughly equivalent to how much value you're going to get out of our solution. And I think you end up with models that look more like that, which is like the more conversations you have, the more you're going to pay us. right? The more data that's involved in those conversations, perhaps, the more you'll pay us, because there's more compute going on to infer things about that customer to do some smarter things. So maybe there's levers that you turn on and off to make the conversation smarter and those are features essentially you pay more or less for, but it's not going to be a per seat kind of basis.

That's very interesting. And maybe we'd take it back to fundamentals for a moment. You guys initiated revenue growth targets for 15% to 25% from fiscal '25 to '27. Right now, the company, I think, is expected to grow less than 10% this year, right? So can you help bridge for the audience kind of how you expect to get back to that kind of 20% midpoint over the next couple of years?

Yes. So there's 2 ways of -- or not 2 ways, 2 facets of the answer. One of the things that we're doing and then the other is like more macro, right? So I'll start with the things that we're doing. What we are very focused on, number one, is scaling and getting the go-to-market machine for Flex and Segment to a scale that is like a humming machine. As we've told everyone in the past, we combined our sales teams in early '22, which was a mistake. And so we've kind of had to unwind that decision and reconstitute the independent selling machines for Flex and for Segment. So that's been a painful mistake. But I think largely, the work is behind us in terms of the foundational work of hiring the sales team, getting butts in seats, enabling them, and now we've got a largely enabled sales team, and now it's a matter of building pipe and closing deals. For comms, what we are focused on is going out there winning design wins. So that's when a customer decides to use Twilio and is going to bake us into a product as well as taking share from other competitors. And in terms of Twilio having leveraged the last the last era of growth to go have the largest customer base, the best balance sheet, the best products and the most revenue, like that puts us in a great position that now we're on the other side of like of the cycle to go win market share because we're the strongest company in our competitive set. So we grow our customer base there. We grow our customer base on both sides. Those are the things that we're working on now in terms of our go-to-market. The flip side is the macro economy. Because as we've talked about, a usage-based model has a great tailwind during a growth period, but unfortunately, has a headwind during the economic not-growth period that we're in, I guess. It's not a recession. The slowdown, whatever the flip side is that's between like awesome times and a recession. And so it's a headwind for us now. For communications, a usage-based model means that like if our customers are just doing less business, less transactions in their businesses, that typically would mean like less communications. So I told the story on one of our earnings calls that I bought the new Zelda video game from Best Buy. And I got a text message that said, hey, your order is ready to come pick it up, right? So as an example, if fewer people are buying things because they have less disposable income, then there will just be fewer text messages sent that because there's fewer things bought. In the same way that people are taking less vacations or whatever, less consumer activity equals less communications. In Segment, there's also a little bit of that dynamic, too, because we monetize it in monthly tracked users, MTUs. And so -- and that includes, by the way, anonymous users. So people who land on the website and are clicking around, right, that's interesting data for a company, because one -- at some point, that might turn into a tracked user and you want to know all the activity they did. But if marketers are spending less on, say, SEM, to drive people to that website, and we know a lot of marketing budgets have been pulled back during last year, well then there's less users to track. And so at renewals, companies have rightsized those contracts to the now monthly tracked users that they're seeing. So that's the headwind we face. So on the macro side, the tailwind is as economic activity picks up, we don't have to do much and you start to see the growth kick in again as more economic activity happens, as marketers spend more, and more people on the website or just more transactions are occurring, you see more in the communications side. And the thing that we are well positioned for is to capitalize on that growth without having to spend more. Because we have reduced the size of the company during this cycle pretty substantially, about 27% overall from our height, through 2 rounds of reductions in force. We've got a really great cost structure in place now, and we can deliver on that growth without having to invest much at all, if anything, into the foundation of the company. And so what you'll see is as that growth kicks in from a macro standpoint, we can drop a lot of gross profit to the bottom line. And that's what's really exciting about the -- what the future holds in terms of things that are in our control and then a little bit of the macro side. And so look, I think that if we hit a really horrible recession, I think that could be an impediment to us to reaching those medium-term targets, but I don't think anyone is calling for that. And so I feel confident that we can hit those medium-term targets.

So from your perspective, like where are we in terms of usage trends? Are you starting to see stability kind of coming out of 2Q relative to where we were in 1Q or even a couple of quarters back? Can you share some context around where you think we are in terms of that?

Yes. In the early stages of this side of the economic cycle, we saw the usage patterns just deviate from where they had been historically. So our predictions about, hey, what do we think our usage is going to be? They weren't materializing because there were fewer transactions going on, fewer like all the macro stuff going on. And when we talked earlier this year about stabilization of volumes, what we -- what that meant is that we weren't seeing degradation. In fact, our models were accurately predicting the usage trends that we were seeing. And so that was the stabilization. And I think that now what we're seeing is both the effect of some -- I wouldn't say like macro is like changing materially day-to-day, but I think generally speaking, you're probably seeing the economists say like, yes, there is a stabilization. It's not kind of the lot of things are up to question like they were maybe a year ago. So there's stabilization going. And then you say our go-to-market efforts of winning market share, winning new customers, et cetera, has the ability to grow our usage volumes and the revenue as a result.

I want to jump in one thing. So on that very topic, Jeff, you're calling for -- at least you expressed optimism that bookings can accelerate in the second half of the year. Is that based on the stability, the economy stability? What are the leading indicators you're seeing in your business that give you confidence that we can show -- exit on a strong note this year?

Yes. And just for clarity, I'm not trying to make any guidance around bookings because I don't think that's something we've ever guided to. But what I will say is like I think we're seeing some of the buying hesitation early in the year as like those dynamics are changing a little bit. I wouldn't say it's like a wildly different market. But I think we're seeing some more positive signals in terms of like early signs of pipe generation and things like that. I think generative AI is certainly a tailwind of this whole thing because despite the fact that customers are still rightly focused on ROI and being good stewards of their capital, they all know that AI is going to change their industries and change their companies. Now exactly how and when, everyone is trying to figure it out. Everyone is trying to say, what are the things I should be investing in that are going to bear fruit? Which are the things that maybe are too early or there's probably going hucksters like in the AI world, just like there were in Bitcoin and everything else. Like every trend brings that. And so they're trying to sort through, okay, what's real and what is maybe speculative. But they know gen AI is going to change. And so that is the nature of every conversation I'm having with customers now. And the way we are positioning, what we do is, look, it is very early in the buying cycles for AI. No doubt about it. I don't think -- I think there are some exceptions, but I think largely speaking, AI is not turning into revenue for software companies yet.

Do you think it will?

Absolutely, because I think the incentive to invest in these use cases is going to be so high. I'll tell you a story. Like we just made it through our migration in our communications side of our business to 10DLC migration, which is an obscure thing that, sadly, investors in Twilio had to understand. But it's like we are migrating from basically a world where every sender on the network was kind of unknown and anonymous to a world where everybody is known, which is actually really good for all of us as consumers in reducing spam and fraud and all sorts of bad things that can happen on the network. This is a painful process because everyone had to register their campaigns to be able to keep sending text messages. So it was a very painful effort. We deployed AI in some of the later stages of getting these campaigns approved that did the work of hundreds of people that were spending months working on it. Did that work in like minutes, right? I mean that's probably 1,000x productivity increase for Twilio. And when you look at these outcomes, you're like, wow. Like it is astounding. And so now you start to think about rolling ahead to other situations that our customers are in. where they are just throwing manual work at problems that could be automated to get a 1,000% increase in productivity. That is -- those use cases, you can't ignore that, right?

Does that lead to contraction in employment or it just makes us more productive and do things faster or quicker and enjoy better quality of life?

I mean, look, the comparison that I typically make rightly or wrongly is to the industrial revolution when you suddenly got an abundance of physical labor, right? And a lot of people said, well, what's going to happen? If you need 100 people to run the firm today and now there's a thing called a tractor, like what's going to happen? Well, clearly, farms are running in a much more efficient way. We're able to feed billions of people that we couldn't feed prior to the industrial revolution. And people have moved on to other professions. And so the farming industry is much more efficient now, you would say, and as well as able to fulfill the needs of humanity much better. And so I think there'll be similar transitions that go on. I don't know exactly where and I'm not going to pretend to have the answers about exactly how it's going to work out, although I suspect that there's a lot of parallels from history that will be analogous in a similar way that you look at how computers made people more productive in their work, and productivity led to a stronger economy, more -- just more economic output from humanity. I think you'll see the same thing here.

Bill McDermott was here yesterday. He's always bullish, but he's been -- for the right reasons, right? He believes that IT spending next year will be faster growth rate than what we saw this year. I guess you know, everything that's gone through. I mean do you have any thought on that? It seems like this whole stability theme is actually -- rates start to come down. We're not economists. You're not, I'm not either. But if things start to stabilize, maybe we unfreeze some of the budgets that have come under so much scrutiny. Do you think there's a possibility that next year could be a better year overall for software?

Well, you know what's interesting is this. Somebody actually asked me in the session just before I came down here, they said, if people are going to invest in AI, where is that budget going to come from? Like who's going to be the loser in that story? And I thought about it for a second, because it actually wasn't something I've necessarily thought about in the past, but it did strike me that I don't think there needs to be a loser in that story. Because if you get as much efficiency gain as I think companies can get with AI, the ROI and the speed of that ROI pays for itself. And so you don't have to have a loser. [indiscernible] investment, but it's just going to pay back so quickly that naturally you get that notion of like efficiency gain and the software industry can return to growth if we're able -- and that's the big if, if we're able to deliver on that efficiency gain. And so I think that there will be certain areas where we'll able to deliver on that pretty quickly, and then other areas will probably take more time.

Yes. So with that, I think, 19 seconds away. Thank you so much, Jeff, for coming to our conference. Really appreciate it. And I hope you have a productive day. And thank you for our clients for your participation in this call.

Thank you. Thanks, everybody.

Third day's the best day. Yes. Let's finish it up on a very high note. Thank you so much.