Including those regarding future events, trends, expectations, projections, and beliefs that may affect our industry or our company, product developments, AI, and potential growth drivers. Such statements or predictions should not be unduly relied upon by investors. Actual events or results may differ materially, and Five9 undertakes no obligation to update the information in such statements. Please refer to our most recent Forms 10-K and 10-Q under the captioned risk factors and elsewhere in Five9's annual and quarterly reports filed with the Securities and Exchange Commission.
Excellent. So just as a brief introduction, I'm Ryan MacWilliams, Mid-Cap Software Analyst here at Barclays covering CloudComm, DevOps, and MarTech. With me today is the Five9 team, Mike Burkland, CEO, and Barry Zwarenstein, CFO. For those in the room and on the webcast, we're not going to be taking questions directly, so if you do have questions, please email me at ryan.macwilliams@barclays.com. I'll get those in. You know, it's been an exciting past couple of years covering CloudComms. You know, this morning they talked about the potential for a super cycle of IPOs. I came over to Barclays in 2021 hoping to get that, and I launched coverage October 2021. And unfortunately, we didn't see that in 2022 and 2023. You know, we've kind of seen the rise of AI in CloudComm, and contact centers definitely have been a big part of that.
But then you also just announced new AI agents and new capabilities for Five9. So we'd just love to hear you start kind of like the lay of the land, where you see Five9 in the contact center space today.
Yeah, absolutely, Ryan. So just a little background. Five9 is providing an AI-powered software platform for some of the largest brands in the world to, we like to say, reimagine the customer experience for their customers. We call it also AI-elevated because AI is a huge component of our platform, and we'll get into that obviously in detail. But to give you a little bit of the big picture, we're about $1 billion in revenue run rate today. Subscription revenue grew 20% year-over-year this last quarter and is accelerated from the prior quarter of 17% growth. And it's an exciting time in our industry, and a lot of it is because of the AI in our platform. We've been in the AI portion of CX for years.
We acquired a company, Inference, about four years ago that really catapulted us to the leader in AI for customer experience. And we also talked on this last earnings call about a lot of AI progress, if you will, in the metrics of our business. 20% of our enterprise net new bookings were AI from a mixed perspective, 40% year-over-year growth in AI revenue. And we'll get into more metrics, but I do think it's safe to say that, look, the investor community has been in a little bit of a show-me mentality for the last couple of years, and it's nice to be able to show the progress.
Yeah, I think your space will definitely be one of the earliest to see actual AI revenue, just how important it is and how helpful it is to the agents. You know, I would just love to hear your view. I know you've talked about ad nauseam, but I hear it a lot from investors of like, how are they just going to get past this AI narrative? And I'm like, well, they are actually seeing AI revenue, so that's a good start. But I mean, from your perspective, I mean, I'm sure you get this even more than me. I guess, how do you think investors come back to Five9 and contact center space?
Yeah, I think it's a great question, Ryan. And again, remember, we're a platform that is a software platform for these large brands to power interactions, whether those interactions are handled by a human agent or they're handled by a bot or a combination of maybe a bot on the front end and eventually a human across all channels, digital and voice. And it's important to understand that we've got to be careful not to just focus on kind of the labor arbitrage opportunity in terms of agents and human agents and the impact of AI. When it comes to a company like Five9, we get more revenue per interaction when it comes to an AI-powered interaction versus a human-handled interaction. So that's one thing. And then in terms of, look, there are only so many CCaaS, Cloud CCaaS vendors, cloud contact center as a service.
It's important to understand there are obviously going to be a lot of players in the AI world of CX. Our customers and the brands that we talk to, they want their AI predominantly for CX. They want it predominantly to be supplied by their contact center vendor. The reason is we have the end-to-end platform across all channels, all interaction types. Those interactions are flowing through our platform in real time. If you think about, you know, I think we'll get on to Agentforce and Salesforce in a second, but look, we've been a great partner of Salesforce's. They've been a great partner of ours, and we are actually, they're leaning in more than ever because they want to fulfill Marc's vision of Agentforce and AI agents. Look, the AI is only as good as the data behind it.
If you don't feed AI the right data and all the contextual data, especially in customer experience, we're all consumers, right? We've all interacted with brands, whether it's a problem or a purchase. You have to have a hyper-personalized experience in this day and age. And the only way that these brands can deliver that, whether it's our AI or it's AI from a third party like Salesforce, you have to have all the contextual data. And the contextual data that is flowing through our platform in real time, whether it's voice or digital, is required for Agentforce to do its job once that product kind of matures, and it has to mature still. And we monetize that opportunity via VoiceStream. So there's a lot to unpack here, but the good news is we're a net winner.
We monetize that VoiceStream feed, so to speak, between real-time interactions that their AI might need. We monetize that on a per-minute basis, and it's significant in terms of the revenue opportunity to Five9.
We'd love to hear more about that. Yeah, I mean, a lot of interest from investors on voice AI. You guys were ahead of the curve with Inference, but just in the voice streaming piece, what do you think that can mean for your business?
I think it could be extremely meaningful. I think in the end of the day, look, most of these large brands have heterogeneous CRM environments across their business, right? They might have Salesforce in a given business unit. They might have Dynamics in another business unit, Zendesk. Homegrown CRM is about 50% of the CRM market. And so the good news for us is we integrate to all those CRM platforms, and most customers want their AI to work consistently across the entire enterprise. And that's why they also come to, again, their CCaaS vendor for their AI solutions.
So can you just give me an example of how VoiceStream would work with Five9? What would be a use case, and how would that link into one of those platforms?
Sure. So again, this could be a third-party bot, for example. We partner with other AI point solution firms, like Cresta is a partner of ours. If a customer chooses to want to use a third-party AI for a certain use case, we have no problem with that. We're always going to do what's best for the customer. That's always been our approach. And the way it would work is if the customer wants to use our platform with Cresta's Agent Assist, for example, we'll integrate via VoiceStream, and we will charge on a per-minute basis for that consumption. And it's typically equating to about, if it's an AI agent in this case, it can be as much as $50 per month in recurring subscription revenue per AI agent to Five9. And actually, we think there's actually upside in that.
If AI agents become more efficient than humans, that $50 might go to $70 or $80. So it's not an insignificant revenue opportunity for us. Now, again, it's early days because most of our customers are using our AI so far, but we have use cases where they're using third parties as well.
Barry, just as that VoiceStream compares to your traditional calling usage, I mean, I know it's still early, but maybe this is better margin or that.
Yeah, it's still very early days. But we're adding a real service over here, providing this contextual data in real time, and that's worth a lot, and yeah, it's too early to give you a margin indication, but it was certainly more like a software than anything else.
Wouldn't you say this is more additive use cases in terms of what you're already doing for your customers? This would be like in addition to the call volumes?
Yeah, I mean, a good way to think about this is, look, there's digital and voice interactions already coming through our classic CCaaS platform. But AI represents self-service in terms of voice bots and chatbots. We obviously have products that do that. But again, if it is a situation where a customer wants to use a third-party bot, we'll integrate via VoiceStream and still monetize that significantly. And then there's all our other AI products. So we have not just AI agents, which we've been talking about, which is self-service, whether it's voice or chat or other digital channels, but we also have AI solutions for human agents, Agent Assist, which include things like guidance cards, real-time guidance cards to help them do their job more effectively, summaries, which summarizes interactions and puts them into the CRM, which is a no-brainer ROI.
We have products like AI Knowledge that basically take unstructured data, so you no longer need a knowledge base. You can just ingest data in unstructured forms. You also have things like AI Insights on our solution, which really helps the business overall diagnose the types of interactions that are flowing through their contact center and identifies automatically. We can turn it on for two weeks or thereabouts for customers, automatically clusters all the interactions by use case and sentiment and impact to the business, and gives us together, our customers and our AI experts, a really good roadmap for where to go first, second, and third in deploying AI to help, again, drive self-service where it should be used, but not over-rotate on automation where, again, it's going to impact the customer experience negatively. Our brand, our customers are really, really trying to drive great CX, right?
A seamless, personalized experience, but leveraging self-service when it should be.
I think that's the misconception so much in contact centers. People think that contact center managers wake up every day and they're like, "I can't wait to cut half my agents." You know what I mean? The idea is like, how can we drive more revenue? How can we provide a better customer experience? That's going to be way more meaningful for your actual business than.
That is the strategic goal is better CX, but let's not lose sight of the fact that there is a real hard dollar ROI on the margin. But this is not the Klarna example. That is not realistic. I'm sorry to keep bringing that up, but it's just a fable. That is not what our customers are trying to do. They're trying to maybe take 3%-7% of their interaction volume and have it handled via truly fully contained self-service. Now, can that increase as GenAI and LLMs get better and better? Yes, but it's going to happen over years, and we are going to get more revenue per interaction for providing either that software, the IVA, DVA, self-service solutions, or we'll get it through VoiceStream if it's a third party. So in either any of those cases, what we care about is interaction volumes.
If AI does its job, interaction volumes will go up between consumers and brands. We've seen it historically in our industry. If you give consumers more and easier ways to interact with a brand, they're going to interact more often. Because we all are consumers. We know sometimes you say, it's not even worth it," right? You know if you've got a really quick self-service way to interact with that brand to do something that's capable of being handled in self-service, you're going to interact more. Five9's business opportunity, our TAM, increases as just the overall interactions increase, whether it's AI handled or human handled.
Yeah, maybe we have to kind of rethink about the seat-based opportunity, the on-prem, the cloud, all that fancy math, and now kind of how does interaction growth play into your revenue? That makes a lot of sense.
That's a great way to think about it.
Barry, I'm going to get to the fun financial questions in just one second. But just to put a button on the AI conversation, I think people underestimate the complexity that goes into the contact center deployments. And the question I get a lot is like, "Okay, all those great AI features you talked about, right? There's other competitors that are offering them. It's like, why is Five9 not going to win those?" And it's like, well, it's pretty complex to put through 10,000 calls in a day, right, on a multi-geo basis. So how does that phone call start the process of you getting the rest of your customers' data?
Yeah, it's actually a great point because, again, it's not always a voice call, but that global voice network that we have that we're delivering at scale with Five9's reliability is absolutely mission-critical and is difficult to do. There's a reason there are only really three cloud contact center providers in the enterprise market at scale. So you've got to understand that that is a, it's kind of like the crown jewel, if you will, of a customer service organization from a technology perspective. You have to have CRM, and you have to have a contact center infrastructure solution like ours. That's never going to change. And again, a lot of these interactions are starting in different channels, right? More digital, but there's always got to be the, and we provide those solutions as well.
But as AI helps power more and more, whether it's digital, whether it's voice, whether it's self-service or human handled, as those interaction volumes go up, they need more of our software. That's the bottom line.
You hear folks, it's like, "Voice AI is becoming more important." I'm like, "Oh, more complexity around inbound and outbound calling?" I mean, I know some good folks for that. Barry, just we were asking everybody on the fourth quarter since the election, we have different answers as a part of this, but have you seen any kind of macro improving? We just went past Cyber Week and holiday season traffic, but any context that you can give us around what you're seeing for the fourth quarter?
Yeah, so what you're really referring to is the seasonal part of our install base. And let me put it in context first with Q3. Back in July, we surveyed formally our biggest enterprise seasonal customers to see what they were doing. And they gave us their best estimates at the time. We also, as anybody who's stayed close to the story knows, we also tracked debit and credit card spending on the theory that 67%, 70% of the economy is consumed, and a lot of it's done on credit and debit cards. And we came to the conclusion with respect to Q3 that we needed to be cautious. And it turned out in the end that 16 of our 17 verticals did in fact come in a little bit better than we had anticipated. The one that didn't is education, our seventh largest vertical.
But overall, it was pretty benign. We took that same prudent approach. So typically post-COVID, we've had a sequential increase of between 2 percentage to 3 percentage points, Q3 to Q4. We prudently kept it at 1% this quarter. And I'm going to leave it at that.
Fair enough. I always encourage people to break news on the fireside. I can usually never get them to.
We resisted the temptation.
I'm trying. I've always done this before.
I'm throwing some lines in the water. Oh man, Billy Ackman comes out. Okay, so when it comes to, you mentioned the mega deals in the past, and we've talked about the ramp timelines. I'd love to just hear kind of where we're at with the two healthcare companies and one large bank as a part of those deal implementations.
Yeah, I would just say our mega deals are right on track. Obviously, there are normal ebbs and flows, but we're really thrilled with the pace of deployment and how much of our revenue they've become. And when it comes to the large financial institution that we won in Q1, we've told everyone very consistently that professional services revenue is happening in 2024, but subscription revenue will not come until 2025.
Is that all start next year?
It'll be a multi-year deployment.
Another large healthcare insurer, I'm thinking of, doesn't have retail locations. That's like a multi-year process as well, right?
Yeah.
Okay, excellent. And then this year, it seems like just given the macro, the dolphins kind of drove your bookings more so than the whales, right? I mean, aside from a really incredible first quarter deal, it just seems like the fishing environment for those big deals wasn't really in the cards this year just because it seems like there's less willingness to have those big on-prem to the cloud conversions, right? Do you think that's fair, and do you think you could maybe see a better environment from that?
Yeah, I wouldn't jump to that conclusion, Ryan. Look, the market is still only 30+% cloud, especially in the large enterprise portion of the market. It's even less than that in terms of cloud penetration. I do think that we've told everybody this, that the timing of these mega deals is lumpy and a little bit unpredictable. The sales cycles are very long. But to remind everyone also to focus on the dolphins because this market's a bell curve. It's like any other distribution of any market with the megas over here in the tail. There are only so many of them.
They got crazy large contact center footprints, but the dolphins in the middle, the $1 million-$5 million deals, there's a lot of them, and they make up a bulk of the TAM, and they make up a bulk of our deal mix and our revenue growth. So this is where people should be focused and consider a mega as kind of gravy on top.
That makes sense. I mean, do you feel like some of these larger contact centers, they've been with the Avaya and Cisco's of the world for 20 years, and sometimes it feels like they're never going to move, right? But do you think AI can start moving the needle from them?
I think it's a real catalyst, and it has been a catalyst. If you look at the ones we've been winning and the ones in our pipeline, a lot of the motivation and catalyst for them to kind of make a decision and get on with it is the ROI that AI brings to the table. And again, it's this business case that is very clear where you're theoretically have a major labor arbitrage opportunity where you're theoretically, again, replacing an agent, a human agent with a bot. And that is, again, that's a theoretical argument in many respects in terms of the numbers that are put in these business cases, but it is a real opportunity for them to maybe grow agents at a lower rate than they would have otherwise, right? So it's opportunity cost, but it's also an opportunity for them to improve customer experience.
You think about it, and you've had a history in this world. You know that if you've got human agents in a contact center, there are a couple of different variables in this. You can keep your agent population low and have your hold times be off the charts because volumes do this. They don't just stay in the same. So they'll allow hold times to get to a pretty poor customer experience level if they want to keep their agent count down. So if you think about what AI can do, it's not just about labor arbitrage. It's also about reducing wait times and improving customer experience because the bots are handling some of that traffic.
Yeah, right. This morning, another vendor was talking about they have a cursing score now with AI, and I was like, I worked in a contact center for student loans. I would have loved to see what my cursing score was. Not me, but some of the inbound calls. Barry, just I have you on stage at the end of the year. I got to ask, you already guided to next year. We'll just hear kind of what built up into that guide.
Yeah. So we felt comfortable endorsing what the street was at. We've done our top-down work. We're still, frankly, in the process of doing the bottoms up. Looking at it as two components. On the backlog side, we ended Q3 with a record backlog, which gives us some contribution coming into next year. If you look at it on the install base side, we've got a situation where in the third quarter, our dollar-based retention rate was 108 LTM. We've told the street that the smart money was betting on a slight decline in the fourth quarter. But that gives you enough to look at what might be happening in the course of 2025. And yeah, I'll stop there.
Perfect. Mike Burkland, I want to touch on one thing you just mentioned. So yeah, I mean, for the lifetime of your business, when a customer moves from a heavy on-prem deployment to the cloud, it would be natural to reduce agents commonly as a part of that, right? Because your platform is making them more efficient. So even though that's happening, that's all net new for your business. And I think one thing that we've heard over the past 10 firesides is when people buy new contracts or new deployments, they're more willing to add the AI features and the agents upfront because it's like, "Hey, let's do it all at once," right? Because you would think like, "Oh, you're already on the cloud for your installed base.
Why don't you just add these agents?" But in reality, it's like, "Let's try something new instead of stopping your way." So are you seeing that?
We are starting to actually see our install base have more appetite and more demand for our AI. We talked about a couple of metrics on the last call. Our AI bookings to our install base in Q3 grew 50% year-over-year, including just, again, we've got nine different SKUs when it comes to AI, and this is across several of those product areas. It's important to understand that we're getting AI traction both within our install base as well as attached to our net new deals. Just to give you an idea, we have a 100% attach rate to new logo wins over $1 million. 100% of them also purchased our AI, and that's for four quarters running. Again, very, very good pull and justification on the new logo win side, but also we're starting to see more demand from our install base.
Part of that is just the AI fog and the distraction factor, I call it, is the learning curve. People are coming up the learning curve. They understand kind of the categories of AI solutions for CX. We're helping them. Part of our role in this whole market is to be the AI experts, our people on the front lines, our out helping our prospects and our install base customers figure out where to go first, second, and third when it comes to deploying AI practically and getting real ROI for it.
So Barry, just on that, so on a like-for-like basis, maybe new customers today versus a year ago or before are adopting more AI solutions. I know it's still early and it's tough to draw too many conclusions from that, but would you say this is driving larger contract values and then maybe healthier unit economics from here with just more AI and more software?
The size of the enterprise deals that have AI in them are larger. Mike just mentioned some aspects of that. The economics of AI are software economics, so it certainly helps improve as well. In fact, when you look at the 130 basis points sequential improvement in gross margins in the third quarter, we listed three factors, the most important being increased revenue against fixed and semi-fixed costs. That'll always be the case. Second one was a fourth quarter of the RIF, or the RIF coming in, excuse me. AI was the third factor there.
And just to quantify that statement, Barry, deals with AI were five times larger than deals without AI in the quarter. And again, some of that self-selection, right? Larger customers purchasing AI as opposed to smaller customers. But it's nice to see that.
I've recently, in the past, Inference attach rates to new deals, like voice AI agents attached to new deals of like 10%-20% as well. So people are feeling more comfortable buying agents on top of human agents at this point.
Yeah, I mean, I think that's really expected to be part of the mix and part of the platform, and again, it's embedded in our platform. They're separate SKUs. We charge separately for them, but they're all embedded solutions as part of our overall platform, and that's really important, but yes, there's appetite for leveraging multiple AI solutions from us in most of these deals.
And I just want to quickly mention, in terms of that last response that I gave, in terms of that gross margin improvement, AI did help, but it wasn't actually, as I think about it now, the third factor.
Okay. And just on the Acqueon acquisition, this was one where it was a little bit of a [audio distortion] for me because I was like, "Five9's great at a lot of these things," right? But what are some of the components that made that unique and important part of your strategy going forward?
Yeah, we got to know Acqueon in many of these large mega deals, frankly. That's where we found a partnership, and we've been reselling their solution. We resold it on that large financial institution deal in Q1. We resold it on deals in Q2, as well as that $4 million ARR deal we talked about in Q3. So we love to acquire companies like that where we're already in the market with them. We know the customer, our joint customers love their solution, and there's a lot of demand for it. And yes, it overlaps a little bit with our Gen One outbound products, but look, we were pretty much a voice-only outbound solution. That market has changed a lot in terms of a lot of the use cases have become unfavorable. And over the years, we've done less and less of that.
It's not really much of our customer base anymore. But omnichannel outbound capabilities, not just voice, but also digital outreach is what Acqueon's known for. And most of these large brands love the fact that they can have this blended inbound and outbound interaction type with their consumers, whether it's appointment reminders via SMS or it's revenue acceleration in other use cases. And our long-term vision for Five9 is to be that interaction, engagement, orchestration engine across the entire customer journey. And what Acqueon allows us to do is start to expand our TAM beyond customer service and into use cases that are touching sales and revenue acceleration as well.
And actually, that goes back to what we talked about earlier, which is kind of more focused on interaction growth, right, instead of classic seats. So that's like more outbound digital interactions and getting better there.
Yeah, that's exactly right.
Just the last one for me, and look, it only took me like four years of covering contact center and working on one to kind of better understand this. But I guess one thing that was, I guess, naive of me was to think in like a Fortune 100 or a large 10,000-seat-plus contact center that it was like one vendor. It's like, man, they won the whole thing, right? But in reality, there's many different components of the CX stack, and there's many different vendors servicing niche use cases or different things. So when people say like, "Oh, Agentforce versus Five9 or whatever," it's like there's going to be pockets that you need people for everywhere. So it's almost like more of a partner still for you guys.
I would even go way more extreme on that statement, Ryan. You have to have CRM, and you have to have contact center to run a customer service organization. You can't get by with just Salesforce, Agentforce. Even after it's fully matured in the AI world, if you have a point solution that's not, again, you cannot survive without that global voice network and the digital contact center platform integrated to that CRM system. So it's not an either/or. It's a, we're both going to win the platform. We are our competitors. I mean, honestly, some of our competitors might win some of the CCaaS platform, but you have to have both CCaaS and CRM. And the question is, what's the mix of AI and who's that coming from?
You've co-existed historically, and that's not going anywhere anytime soon. But for those in the room, if you have any questions, we can definitely get this over to Five9. But Mike and Barry, thanks again for being here.
[crosstalk] Thanks so much, man.
Thanks, guys.