My associates were kind enough to host this for me because it was a week after the birth of my son, so I was not going to make it.
Congratulations.
But we celebrated a year for Colin, so things are good. But it's great to be back. So I'm Ryan MacWilliams, mid-cap software analyst here at Barclays covering CloudCom, DevOps, and MarTech. With me today is Twilio CEO, Khozema Shipchandler. And for those who are in the room or on the webcast, please email me at ryan.macwilliams@barclays.com as we won't be taking questions direct. But Khozema, thanks for being here. How are things?
Things are great.
Yeah, thanks for having me.
I feel like we've hopefully turned a corner for mid-cap software because I launched coverage in 2021, and October 2021, and 2022, 2023 were some interesting years. You guys saw revenue re-accelerate in the most recent quarter in your comms business. How can you just kind of give us kind of a breakdown of how some of the changes you guys have made in your business over the last year, year and a half has kind of led to where we are today?
Yeah, I mean, I can't comment on mid-cap software broadly, but I think for Twilio, it really just starts with running the company better. And I think as I think about the company and as I start to think about 2025, what we want to be known as is a company that's run really well, but that's also highly innovative. And I think over the course of 2024, what we really try to focus on, Aidan and I in particular with our management team, is showing financial discipline, running the company well, so showing a lot more operating rigor. And then as we focus our innovation bets, being really focused about them. So instead of focusing on 10 things all at the same time, all with equal intensity, which was easier a few years ago when money was free, be really deliberate about it.
Three to five things, get customer validation, be smart about the additional money that we attach to those projects. And I think we're starting to see that kind of play out. In our own company, volume sort of stabilized over the last couple of quarters. It inflected a little bit in Q3, which is certainly very exciting for us on the back of a lot of improvements that we made in profitability. And I think now we sort of have the size and shape of the P&L that we like. We'd obviously like to continue to grow at double digits. That's something we certainly aspire for. We see a lot more opportunity in the rest of our P&L in terms of free cash flow and operating leverage. And that's how we're going to run the play.
And can you just talk about some of those three key focus areas? I mean, you guys divested some pieces of your business, refocused on certain products. I guess definitely more ROI-based, trying to drive value out of the business. But how would you describe, OK, the entire Twilio employee base is aligned for these three principles or three go-to-market products?
Yeah, I mean, I think it starts with communications, right, because that is by far and away the biggest piece of the business. It's nearly 94% of our revenue. And so to pretend otherwise would be silly, especially given the fact that we're the market leader and that we have a lot of strength in that business. And I think for us, and you mentioned some of these products that we moved away from, for example, it was important for us, both with our customers, but also with our employees, to reestablish ourselves as a platform company. We are fundamentally a platform company. We operate at the infrastructure layer, but we're not going to get into some of these more application-oriented areas that we dabbled in in the recent past. They hadn't been a success. In some cases, we ended up competing with some of our most valuable customers.
And that's just not a winning hand. So I think really being circumspect about who we are. I think then on top of that, adding innovative capabilities to the core communications business that you've got. So when you're the market leader, when you're very disciplined on price, when you're winning with your technology, I think what customers expect is that you're going to continue to add to your existing technical base. So for us, adding channels, adding capabilities like Verify to our messaging business, or adding Voice Intelligence to our voice business, these are things that they create more ROI for our customers. They make them more attractive. And then for us and our investors, they're more margin-accretive. They create more durability on the revenue line. And that's attractive as well.
And then I'd say finally, we've kind of had a conversation about Segment with investors over the last few years. I think it's important to remember, first of all, it's a pretty small piece of the business, less than 7%. But beyond that, it's really the data capabilities inside that business that make it special, right? So the world that we see to kind of now come full circle to your question, what we tell employees is that it's communications plus contextual data, which get unlocked by Segment, plus AI that yields this really awesome customer-to-consumer experience over time. And we think the company and our share owners are going to be beneficiaries of that.
Excellent. And you've definitely seen that in your most recent quarter. You talked about your comms business messaging growth, inflecting the dollar add kind of hark back to a few years ago in terms of the dollars added between 2Q and 3Q. Some of it could be macro. Some of it, I think, also is part and parcel of some of the go-to-market changes that you made. I'm sure it's tough to disaggregate those two, but I guess how would you break down the components of the revenue stream that you saw in the third quarter?
Yeah, I don't know if it's that much macro, to be honest. I mean, I think our planning assumptions are always sort of a neutral macroeconomic environment. I think it's really hard to call the ball on exactly where we stand on macro more broadly, which is why we plan that way. I think what we did see in our business is that if you look across a number of different industry verticals, there was more or less strength in each of those verticals. I think the only real soft spot, and we've called this out now for probably two years straight, is social messaging. That's been a little bit weak. But I think we've lapped that now a few times. So it's not really affecting growth rates or anything. It's just an area of weakness.
I think beyond that, we're very excited about strength that we've seen in some ways, two sort of ends of our business, one being self-serve. That's always kind of been the bread and butter of the business. Developers attach themselves to this platform. They get going. We've done a lot of work with tooling, with AI, to make that much simpler so that customers can get activated faster. I think it's certainly our goal that any AI startup, the first brand that they think about is Twilio when they get up and running. And I think we've seen some success there. A lot of these AI startups, they're going vertical. They're going with voice virtual agents. They're going to ride on our voice APIs, and that's pretty attractive.
But also on the flip side, as it relates to partners who we think can be kind of leverage distribution for us, really good strength with ISVs who they have a much broader customer base. They allow us to kind of both sell into those customers and then allow our products to get out in the marketplace. But also increasingly, they're using our data capabilities, and we can also use go-to-market partnerships with them to package up solutions to bring to customers.
I definitely want to get voice AI a little later. Just on that point, we've tracked a number of smaller voice AI startups that are trying to do more generative AI there. When we see who's using their underlying platform to provide the voice, it's you guys.
Right.
Is that just a part and parcel of the strategy change you talked about? Is maybe in the past you would have like, we want to be the voice AI platform, but now it's like, look, we'll also service the app vendors that are trying to attack this themselves?
Yeah, I mean, these vertical things, they're really awesome and exciting, and we want to support all of them. But they're fundamentally applications, right? And we're not an application company. I was being very deliberate in saying we want to be a platform company. I think when we've dabbled at that application layer, it's either put us on the wrong side of some of our really important customers, or it's made us pursue a thing that we don't have the DNA for. And so I think it's important to know what you're good at. And in that respect, we're going to stay at the platform level.
Excellent. And I'm going to get back to that. But just on Cyber Week, you guys put out some stats around how things went this year for Twilio. Pretty strong growth versus last year. I mean, last year it seemed like it was roughly flat for total number of messages. And this year it seemed like over 20% growth. I'm not trying to make you comment on anything in the interquarter, but as I look at these.
I'm sure you're not.
No, no, of course not. But you can break news if you want on the far side. But as it comes to when I look at, wow, and Salesforce, Holiday Insights, a bunch of other vendors, we saw a strong mobile and overall messaging Cyber Week period. Should we take that as a healthier environment for messaging overall, or is that a decision for we're going to put more marketing content around this week period separate from an overall macro?
Yeah, I mean, I think the marketing of it is mostly with respect to there's this season in which there's been consumer strength. There's been a lot of activity through digital platforms, and just look at the sheer scale of what's going on inside of this company and our ability to continue scaling at ever-increasing and accelerating volumes with customers. What most people don't realize in this business is that when you look at CPaaS broadly in the U.S., but also in other markets, but just to stick with the U.S. for a second, telco supply is actually limited, and the demand is always increasing, right, so your ability to do one thing in a microsecond, maybe delay the next thing for four seconds, and perhaps even deliver the next thing in 10 minutes, that matters, right?
So your ability to kind of throttle through that and keep customers delighted no matter what it is that they're sending is really, really important. I think what we saw is there's strength with the consumer. I think a lot of that already went into our planning. We'd seen some spikiness. And so when we provided guidance for Q4, a little bit of that was baked in. But I think it really just goes to the strength and scale of our platform.
So one thing we saw across the Cyber Week messaging patterns was more multi-channel messaging. And kind of what you just described in that example was a more complex marketing campaign, like this minute, this hour, the next day, hit them with multiple different messages, right? Do you think your platform approach is resonating with folks as they try to get more complex, especially with AI across different channels?
Absolutely. And I mean, I think what's also interesting about it is, as you kind of consider how does AI maybe figure into communications alone, is what I think most people don't realize. When we say contextual data, I think most people think we're only and exclusively referring to Segment. The reality is that we literally have trillions of data points and communications events, right? And in that data is a lot of signal about, OK, well, what does the customer actually like? What do they actually pick up? When do they like to be reached? What's the medium, right? And so you referenced multiple channels.
And so I think what the multiple channels allow for is A, a little bit of experimentation, especially with the newer stuff, but B, with all of the things that we know already, with one customer, what might be perfect is hit them with an SMS and then immediately after that, hit them with a voice call because the voice pickup rate will accelerate so much more significantly if you do it that way. And even with the added cost, the ROI pays off. Or with another customer, you know what? They're never going to pick up an email. They're always going to pick up an SMS. We happen to know that because we're sitting on all that data. And so target the customer that way. Or in some instances, it's just at this time of day. Never bother them at this time of day because you won't get them.
But at this time of day, send them anything, and they'll engage with you. So there's a lot of value in our customer's ability to reach their consumers just in those data points. And then when you layer in the other bits of contextual data, like demographics, buying patterns, web behavior, that's when it becomes super exciting.
And that time of day messaging sounds intuitive as an investor. Everyone does this, but that's still so early for many different.
It's so early.
Many different brands.
It's so early.
I still remember I was at this app player investor event, and they were like, OK, if you're an email, if you send emails, we have three tables. This is the expert table, medium table, beginner table. And it was like 25 people. And 15 went to expert for email, five went to medium, five went to beginner. And then the next session was SMS. And that same 25 people, two people went to expert. And then the rest went to either medium or beginner. And I'm like, email didn't come out yesterday. Or sorry, SMS didn't come out yesterday. What's going on here? And it just is like, look, people are still really new to incorporating even SMS or multi-channel messaging in their platforms.
I think it also gives you a sense of why the TAM is so relevant and underpenetrated, right? There's still a lot of greenfield opportunity here, even in the core channels.
And I would just look at the Twilio Black Friday results there. Just voice grew stronger than I expected, like almost 20%.
I think part of the reason for that, I don't know if we're going to talk about AI more, but you mentioned the voice AI thing. But I think it's the most natural way today. I think it'll change. But I think today it's the most natural way to interact with a virtual agent, right? Because you still, just like you and I are talking right now, this would be hard over text, right?
Yeah.
Possible. And these things will happen over time, but it would be hard. And you don't get a sense of the texture of the conversation. You don't get a sense of the emotive characteristics of a conversation. And all of those pop. And what I think is exciting about voice AI in particular is not only does the actual flat data get stored, all of the other characteristics get stored as well. And so if you knew that someone was super angry during one particular interaction, the next time around when you're driving that interaction, and if it happens to be in a problem resolution environment, you might know how to train that agent better to be able to approach that customer the next time around. So there's a lot going on there that I think is really interesting.
Cool. That's a great point, and I can feel the investors are like, Ryan, you need to get to the AI point for Twilio. Just before I get there, and by the way, if we did a chat-based fireside, I think that's how I get replaced by AI.
We don't want that.
Yeah, hopefully it's a little longer than today, but just your targets for next year. It was interesting you guys provided that in your most recent earnings. What kind of helped you build up to those targets? And I'm sure it's part and parcel of a lot of the stuff we talked about, but I guess why now to guide to next year?
Yeah, I mean, I think we're doing this investor day that is on the 23rd of January. Kind of going into that, we wanted to give folks a sense of like, OK, well, what are kind of the broad strokes of what could play out here? I think you've heard Aidan say a number of times that we see ongoing operating leverage in this business, but we hadn't really provided any commentary on the growth characteristics. And so I think it felt like the right time. It's a usage-based business, so we never have perfect visibility. That's sort of the good and the bad of the business in some ways, but we know who we are. And I think on the flip side, we do spend a lot of time talking to our top 250 customers or so.
We do a lot of triangulation in terms of what's sort of the usage that we're seeing. And I think importantly, we're planning for a neutral macro. So when you add all that up, it felt like we'd said enough on margins. We hadn't said much on growth. I think going into investor day and at least having that landed before we got there and provided a broader framework, that felt important to us.
It makes sense as your business becomes more enterprise and ISV becomes a bigger part of your business that you've better insight and outlook over the next year instead of a quarterly basis in terms of usage. Just as a part of that guide, any context or any way we think about revenue contribution between Segment and the comms business building?
Yeah, we're not going to guide down to that level. I mean, I think even today we don't really break apart the business in that way. And look, again, at nearly 94% of the revenue of the business, I think you should imagine that the vast majority of the growth characteristics come from comms, and Segment will contribute over time.
Totally fair. Had to try. Just on the OpenAI partnership, getting a lot of calls about that, I guess I would start with like love to hear about their Realtime API product and how you guys facilitate developers wanting to add that into their products.
Yeah, I mean, first of all, really excited to be partnered with OpenAI. I mean, obviously, they're a company that's kind of really captured the zeitgeist of the moment. I don't know if you've tried the demo or not. It's cool, right?
Totally.
It really gives you a great unlock in terms of what can be done. I think what's exciting for us is that if you're a company, all of a sudden you can look at this technology and credibly say, you know what? My costs are about to go down, and my revenue is probably about to go up. The cost piece is obvious, OK? The revenue piece, less so. If you think about the revenue piece and you think about a customer service environment or a food delivery company or what have you, the CSAT, or excuse me, the metric that matters most is TAT, right? How fast am I completing this call? Now you've got an AI agent who can literally spend forever on that. That's not ideal, but can spend a long period of time on that call.
You have certainty that whatever the problem is that needs to get resolved is going to get resolved. I think more interestingly, for a lot of our customers that we've talked about it with, you also get revenue uplift, right? Using that food delivery example, it turns out, and this is probably not news to you, but soda is an extraordinary markup, right? Very, very high premium product when your core goods are actually relatively low or stable margins. Every time that you can accomplish that upsell because you have five extra seconds, that's attractive, right? These companies are able to drive more revenue. What's even more interesting about it for us is that if you consider our comm stack and what we're doing with Segment, we can facilitate all of that.
We can bring in a partner to be able to deliver on the voice agent experience. But what's less well known is that we can also then take all of the attributes of that call, pump them into our Voice Intelligence. That Voice Intelligence will store that information in a Unified Profile. Whatever other interactions the customer has with that business will also get stored, will also be used to train the model for the next time around. And oh, by the way, just to make it delightful, we'll send the customer an email after the fact, thanking them for their business, knowing exactly what it is that they did with us most recently, and then pushing them a coupon for the next time around. So you're able to complete the whole value cycle that's always been imagined, is now possible through technology. Pretty cool.
Yes, for sure, and it's reflective and it builds on itself with more usage. I mean, what we took away from it when we were utilizing it was it seems more approachable for smaller brands. I don't have to hire a system integrator and spend $50,000- $100,000. We can try this right now, right? Are you seeing enough initial test cases, or have you seen a lot of initial interest, or maybe tokens need to come down in price before people will start to utilize them more?
No, we've seen good uptake, I think, is the answer to the question. I think here's the one thing that I think customers are going to think about, OK, is it's still their proprietary data, right? And so the reason that we kind of hang our hats on communications plus contextual data is that it is virtually impossible for me to imagine a day in which our customers are saying to one of the AI companies, you know what? Take all my data, right, and train up your models and then disintermediate me down the road. Right? I mean, it's illogical, right? So I think that they want their data in a warehouse. Every single one of these guys uses a warehouse, and they want one copy of that data, right, so that they can satisfy their CSOs. And we have the same requirements inside our company, by the way.
On top of that, what they want, though, is interoperability with that warehouse where the AI workloads can run on top of the warehouse without that data having to leave. So that way, the customer's data is protected that they have about their consumers. It doesn't matter to us, right, because we're just operating in whatever the environment or sandbox that they want us to operate in. But my point is that I think the friction, if any, will be where does the data go? And so I think that's why it's so important to us to kind of stand behind first-party data, stand behind the warehouses, and say, look, this is maybe a different way to architect it in which your data doesn't leave.
No, that's a fair point. I have received some investor questions around the integration you announced last year with OpenAI and then the new partnership for OpenAI's Realtime API. Is there any differences there, or is that similar?
I would say the partnership initially is kind of more like how do we work together better. And this one is just kind of the natural evolution of it where virtual agents have taken off. The AI capabilities inside them have taken off. They've obviously done unbelievable work in terms of developing their tech. It has to be delivered through a voice API at some level. And so it just felt like a really natural thing to do with them. They're an awesome partner, awesome company. We are going to stay open with it, and we're going to want to be able to work with a variety of folks. And I suspect they may want to as well.
When it comes to your AI strategy over the next 12- 24 months, it makes sense that you say voice is a very intuitive channel for people to utilize AI right away and/or interact with it as a consumer. Do you feel like it starts to show up in other pieces of your business?
No-brainer. It'll go to text. I mean, I think there's no question that that's the next natural step. I think what's actually more natural is multi-channel communications through AI. I think there'll be some interactions that are best handled through voice because of the way that it works. I think for a lot of others, text will be just fine. And then I think if you want it cheaper or if there's some different time dimension, I think email will be just fine, but will also be AI-driven. And the way that we intend to do it is that for us, the core of it is the Unified Profile. If you have all of this information, and only with all of that information, can you keep the cost down? Can you create the richness and the experience to be able to drive the AI workload?
And then there has to be a communications event somewhere in the middle of it, right?
Yeah. I mean, that's how you get the ROI. I mean, when it comes to that, your CPaaS competitors would also say, we're going to go after the same opportunity, right, in servicing these AI workloads. Is Segment a differentiator there as that system record or customer profile to add that extra layer of contextualization with Twilio?
100%. I mean, I'm not at all worried about our other CPaaS competitors. I mean, I think we're already the market leader. We win on our technology. We already have kind of we're a share taker. We have an accelerated product roadmap relative to those guys in terms of what we've announced. But absolutely, having a contextual data asset inside the company that can allow us to create these really interesting AI experiences, huge advantage, huge advantage.
I love that the better, stronger focus on the ISV program and communication behind that. Is there a way you could frame for investors or in the past, have you talked about the margin structure of that ISV partnership and how much revenue that correlates to your business today?
Yeah, I'm not going to disclose that last piece. We'll provide some detail on that, though, at our investor day. I think the trick is with ISVs, but I'll also expand it to all partners. I think partnerships has not been an area of strength for the company, OK? And there's no good reason for that, honestly, right? So let's make it a strength, OK? And I think that's the opportunity that investors should think about. I think with ISVs, they've been awesome because at some level they've got to ride on our rails, OK? But how do we make each other successful together, right? So I think we'd love to transform these into go-to-market relationships where we're taking their kit to market, they're taking our kit to market. We're not competing with each other, so that's straightforward. They're not developing our stack.
We're not developing our stack, their stack. We exited Engage Premier, as you know, and I think that puts us each on the right side of each other, so I'd like to expand our distribution that way, and I think they'd like that as well, and I just got an email from one of them, actually, this morning about that, and then I think the same can be said, though, about the boutique SIs, the GSIs, the hyperscalers. It's just an area of focus that we've not put historically, and I think on the come, that should be exciting for investors.
This is a good time for this because as these application vendors or ISV partners are experimenting with all these new use cases or omnichannels, right, you can be the one-stop shop for them as part of that. I have probably published more on this than some of my peers, and maybe I'm too excited about this. I just think the RCS technology and the use cases that are available to text or marketing use cases that could be unlocked with two-way messaging with RCS are really cool. Maybe I'm doing too much too soon. Since it's been on the new iOS 18, have you started to see more interest from customers on RCS? Do we need carriers to really get there first in terms of backing it before enterprises want to use it? Or how does that rollout work?
I'm kind of neutral to positive on it. We're certainly ready for it. I think Google is a fantastic partner. But I mean, just I'll ask you a question. How many RCS messages do you see?
Not yet.
I mean, I think that's part of it. I also think the use cases are going to be a little bit limited in the utility. You don't need RCS for 2FA, right, to take maybe a more basic thing. I think for a lot of marketing, it's probably overkill. That said, it's pretty slick in a variety of other instances, and I do think there's some interesting opportunities there. The carousels are super cool.
Yeah, absolutely.
But I think it's neutral to positive for us. That's how we kind of view it. We want it to be successful. The economics are about the same as what we do. But a lot's been written about different channels over the years. And I think there's a reason SMS is the universal standard.
Absolutely. Overkill is underrated, in my opinion. But when it comes to, you've seen RCS take off internationally in India, so there is some precedent for this time. But I guess, yeah, we'll have to see how it comes through in the U.S. I just have to touch on this because it is an election year. I know it's less of a focus from Twilio, but I've had some investors are like, it probably did even more and it's going to contribute more in 4Q. But I guess we'll love just kind of a refresher on how you view election revenue within the Twilio ecosystem.
Again, I'll ask you a question, and then I'll answer it.
Yeah, sure.
Do you like receiving political text messages?
Absolutely not. I have a PA phone number, so I was literally getting 10 a day.
Yeah. So we know that. And so we backed away from that business a couple of years ago. So in 2022, it's not zero, OK, but in 2022, we kind of made a conscious decision to really raise the bar on what met our acceptable use policy. And in so doing, a lot of providers walked away from us. And then on top of that, there's opt-in consumer consents. And we thought those were actual things you had to follow, right? And it turns out others didn't. But we felt like that was the right thing. We felt like it was being on the right side of the consumer. And so we more or less backed away. We have the data. We broke it out in our most recent call.
What the contribution was, I'd say more or less in the ballpark in Q4, given the continuation of the election cycle. But it's just not a material contributor to our business.
I will say thank you because I checked 100 of my text messages that came from to buy an Elon Musk hat or something. None of them came from Twilio. So really appreciate that.
Great.
Khozema, thanks for being here. And if you have any questions for the Twilio team, happy to send them over.
OK, great.