Business. We had to kind of retool and change our sales structure to be able to make sure that we're hiring the right AI experts, that we're showing up with our customers and explaining, you know, the real ROI of specific use cases. We made that change in Q2. Actually, in Q1, we made some major changes to the changes we made in Q2, which were more on the new logo side. In Q1, we made some changes to how we focus on our install base. If you look at our Q4 results, we had the largest install base bookings quarter we've had in eight quarters. We kind of had almost three quarters of a year of those changes in place. Q3 had a strong quarter. Certainly, Q4, which we just talked about.
It is kind of that was the navigation of 2024, which kind of has set us up for 2025.
Yeah. I can talk about the guidance piece of it, Tom. If you think about 2024, there are two key factors that sort of took place from a revenue perspective. We had our largest customer who was ramping, and it's a multi-year journey. The contribution to revenue is typically much larger on the back part of that ramp cycle versus the beginning. That customer has significant growth throughout the year. That was one. Number two is in Q4 specifically, our seasonal verticals, consumer and healthcare, actually grew much stronger than what we saw a year ago and stronger than what we anticipated. While those are positive tailwinds last year, now coming into 2025, that does create a tough comparison. By that, what I mean is I'll talk about seasonality first.
You're gonna have a downtick in Q1 that's stronger than what we saw a year ago. Also, looking further out in 2025, given all the uncertainty around macro, given the fact that data points like retail sales being at the lowest point in the last two years in January, we're assuming that the seasonal uptick is more muted in the second part of 2025. That's one thing. The second thing is on that point of the largest customer ramping last year, we do have a strong backlog of customers that we have great visibility into for 2025. One of our largest customers will start ramping in 2025. However, it's that multi-year journey I talked about. The beginning part's always smaller, so you create a tough comparison.
Having said all that, what I will emphasize is the fact that our guidance philosophy is prudent, and it's consistently been that way. I also wanna emphasize that in 2025, we're not in this environment, we're not expecting big beats throughout the year.
Right. That's interesting. Can you just, you know, this consumer part that, you know, and we've been talking about credit card data.
Mm-hmm.
Barry, for a while. You have to take that over, Barry. Walk us through in terms of, like, that second half guide that sounds like it's a little bit more conservative. I don't wanna misspeak in terms of maybe 'cause this is going into the third year, right? This would be the third year in a row that the consumer was a headwind, starting with the 4 Q23 that we started talking about this. How, how, you know, where in terms of the relative size of the business is this consumer part now?
It's a third biggest vertical. Still third. It's important. It's more seasonal than most. The key thing for us is that data that you referred to from JP Morgan, it tracks our internal data very, very closely. And the data is clear. When you look at Q3, I'm gonna give you by month the nominal year-over-year growth. So one, one, two in July, August, and September of this past year. And then three, four, four. Much stronger in the fourth quarter than we expected and I think many expected. The first real growth that we've had in that consumer discretionary spending, for since the beginning of 2023 when it was, and get this, January, February, March of 2023, those numbers were 12, 8, 5. That's growth. That's when you see 20% overall corporate growth. That's not what we had right now.
As Bryan said, and I just wanted to emphasize this, when we set the guidance for 2025, we were very cognizant, as Bryan said, worst retail sales in almost two years, Walmart numbers. In retrospect, as the weeks have unfolded, we're very happy we took the stance we did in terms of that guidance.
Interesting. Thank you for that color, Barry. You know, we're lucky you mentioned Andy, AI, and AI monetization. We have the CTO and the head of AI, Jonathan Rosenberg, here from Five9. Could you just maybe walk us through where you kind of see AI monetization, say, even out a year, maybe even two years? 'Cause right now, Five9 was early with IVA, the intelligent virtual assistant, and obviously was quick to capitalize. I think it was in the first half of 2023 on GenAI services like Agent Assist and Summaries. Where, you know, but what, there's some really large opportunities that you laid out in the most recent call that are data-driven, like, related to, I think it's the, is it Voice Stream, transcription?
Voice transcription.
Just, you know, walk us through where, how this business is gonna evolve for Five9 and how large it can become. Before, if you don't mind.
Yeah.
Before Jonathan jumps in and gives you the wonderful color around our AI story, I wanna level set it. I know you want to talk about the future, but the future starts now. The just completed quarter, we are talking about considerable momentum. If you talk about revenue, so a little bit backward looking, we had revenue growth of 46% in our 10 AI SKUs and acceleration from the 40% that we had in Q3. It is now taking, there's no hand waving about this. 9% of our total enterprise subscription revenue comes from those 10 AI SKUs. Looking at bookings as a sort of precursor to what might happen down the road, we had 50% growth in our enterprise new logos. We had more than 20% of our enterprise bookings was from AIs. We had 100% attach rate on a million dollar plus customers again.
If you look at the install base, we had coincidentally also a 50% increase year over year in terms of AI SKUs. So if that's an AI loser, I'd hate to see what an AI winner looks like.
There's a lot of name-calling in 2023, Barry. I guess he's taken some of the jelly out of your donut there, Jonathan.
No, no.
Maybe AI will be a higher %.
I think it's important for Jonathan to talk about sort of ROI.
Yeah.
You know, that what are the ingredients of why are we winning and what's the future look like?
Yeah. 'Cause I think that's the main thing to think about is which segments are gonna see the biggest growth and success in AI, right? And how are vendors like us positioned against hyperscalers, against point AI vendors, against CRMs in this emerging battlefield. Here's sort of the calculus that I think about how this adds up. No matter who does it, if you want a chatbot, you need four things. Now let's start with the chatbot. Like, or, or sorry, an AI agent that's over voice and digital. If you want an AI agent, you need four things. Okay? You need four things to build one of those. Anyone needs all four of them, not just us, anyone. Okay? You need a large language model. Great. They're a commodity, right? I'll talk about that a little bit more in a moment.
The second thing you need, and this is really important, but I think underappreciated in many communities, is you need the communication channel, the ability to send voice calls back and forth with the consumer for a voice bot, the ability to send them chat messages on the website for a web chat user, the ability to reach them on SMS on their mobile phone or if they're a WhatsApp user to connect them over WhatsApp. In our industry, we call those the channels. You have to have that. Like, if you don't have that, there's no customer in the customer service part of the equation. Like, that is an essential ingredient in doing this. The third ingredient that you have to have is you have to have what we call contextual data.
That's stuff like the customer's account balance if you wanna build a chatbot to answer questions about their account, right? That stuff often lives spread over tons of systems, that are different databases and third-party SaaS products that have been built in enterprises over years and years. The fourth ingredient is the history of previous conversations between the customer and the consumer and the brand. That's an essential ingredient for personalization, right? This is the reason why people hate these bots today is they don't remember that you were there like yesterday and you had a conversation with an agent about this, that, or the other thing. People really gonna use these things, it has to know that stuff. You have to have those four things.
Every, and so when I look at those things, the CCaaS vendors and Five9 in particular, we own outright two of them. Like, we are the channel source. That's the gatekeeper. We are, our platform is the gatekeeper in essence of the communication in and out to consumers. That means that we have dir, we can integrate it directly for differentiated product value, or we monetize it when third parties sit on top of our platform. We have products. You mentioned Voice Stream and Transcript Stream. Those are products that when customers choose third-party AI vendors for an AI agent, we monetize by selling those SKUs to the customer. We ourselves, of course, build a powerful AI agent, with the same tech. We own the channel. We own the conversation history.
That's literally the core data here, the communications between a customer and a brand and all the past chats and voice transcripts. Our platform is a system of record for that. That gives us direct access to it that we can use in our product or monetize third-party access. For integrating third-party data, that's what our platform has done for 20 years, and we're really good at that. We have a services team that Andy's team runs. We have great technology for it. LLMs, they become commodity. That's good because Five9 picked when I started here about six years ago, everyone was like, "Oh, you gotta build your own speech recognition model from scratch. That's what the market wants. Go build it." You know, that's the only differentiator. I'm like, "No. This stuff's gonna get commoditized.
The hyperscaler's gonna build this thing out. It's a race to zero on price. It's madness to build something when the race to zero is on. Thank goodness we made that decision six years ago, and we continued to make it when LLMs came out. It's been completely vindicated. You can't, you can't even build your own LLM. Only the massive hyperscalers can even do it. What everyone is doing now is what we did all along, which is you plug into these things. They're commodity. As their price points decline, our prices go down and we get to do things we couldn't do before. Our engine-agnostic strategy, as we call it, has proven to be very prescient and has given us a step ahead in our ability to ride the economic wave to zero on those models.
That's the calculus on this thing. That's why I think we're in a pretty good position. If you don't believe it, he just gave you some pretty compelling numbers that show, like, it's working. Is there. That's my story.
Back maybe to the higher level way we look at this, you know, there's been success with IVA and some GenAI, GenAI solutions. Is there going to be a mixed shift, maybe go looking out a year or two to more of a consumption-based model related to that data, the data streams that you control?
Yeah. I can go ahead and take that one. If you look at, I think there's two things that are gonna sort of accelerate our AI business is, so we have IVA, right? We have 10 of these SKUs. I think what you're gonna see is we have a solution like AI Insights, which we can talk about. We obviously have our IVA. We have Agent Assist. You start to bring these together from a bundling perspective, package them in a way from a consumption perspective. Let's be clear, our 10 AI SKUs today are all either consumption or capacity-based, right? Different than the seat-based model that we have. We can talk about that as well. I think you're gonna see a lot of the bundling of these things come together and how they integrate.
Back to the monetization point, we get about $40-$50 per AI agent when you talk about the Voice Stream and the Transcript Stream. It is still a, you know, obviously we wanna land the CCaaS platform. Ultimately, you know, we're gonna continue to win, win our fair share of our AI solutions. Certainly, you know, the CRMs of the world, Salesforce, they announced Agentf orce. They're gonna have some success. I think we, we can talk about that when we talk about partnerships. I think that's where a lot of the future AI is gonna come from, is the coming together of these applications.
Yeah. Also on this too, like, if I can, like, I'd also characterize this as like big misconception number two about AI in the contact center is it's either you have a human agent or if you have AI, there's an AI agent and that's your AI story. It's from making self-service AI agents. False. There is a lot, we have 10 SKUs that do AI things associated with the contact center. Some of them apply when there's a human agent. Some of them don't. Like this thing he just mentioned, AI Insights, that's a product we just went generally available on. It's amazing. By the way, it works whether or not there's a human agent or an AI agent in the story. It analyzes the conversations. Remember I talked about our channel as the, our platform is an egress point for all these communication channels.
When you have that data, we use GenAI to analyze it. As long as there's people talking to somebody, whether it's chat or voice or an AI or a human, doesn't matter. As long as there's stuff happening, we have value to add on that platform.
Most importantly, if you look at any customer journey, you're gonna go back and forth potentially from, you might start with a chatbot and then escalate to a human.
Yeah.
To be able to go back and forth between platforms, you know, that's hard to do. It's hard to do things.
It's one of our strengths too.
So.
Yeah. I mean, we've been talking about the, the medium of voice is not going away. We are talking right here.
You mentioned partners. That's where I inserted. I was gonna go, Andy. Talk to, and Jonathan mentioned system of record. That's a strong word in the software world. What does Five9's relationship look like in terms of that term, system of record, when you go to market with ServiceNow and Salesforce of the world?
Yeah. You can add on, Jonathan, but one of the things I think Mike mentioned is, there's system of record, right? I think maybe the term we've been starting to say is system of interaction history, right? When you look at the conversations that happen within a contact center or a sales contact center, we have all of that history, right? From the voice to all the different channels that Jonathan talked about. That's what we are. You look at the classic term of system of record, right? It's the knowledge of the customer information around them. The key is you need both of that data to be able to deliver strong AI, right?
When you look at who's got the upper hand, you know, I think in order to build any of these things, you have to have both sides of it. I think that'll continue to be the case even when, you know, some of the CRMs start to have more success on some of the areas that we're at. You're still gonna have to have that data.
Yeah. On the engineering side, we use the term system of record to just mean the platform that produces the data. If you produce the data, you own the data, right? It's that simple, right?
Yep.
The CRMs, they're the system of record for like your account number and, you know, your last product you purchased and, you know, how hot is your leak, you know, how much did you buy last year, right? Their systems produce that data. They collect it and they produce it and they analyze it. There's other systems like a billing system, you know, or a provisioning system. That's a system of record for like what your cable modem is. Five9 is a system of record for the conversation data, the interaction history. That is one of the critical ingredients for building one of these AI agents. That is why our platform, the CCaaS platforms, are in a really strong position here. Maybe just one more on, or maybe a couple more on, on the partners.
We talked about, I think on the call about, you know, maybe from a using the partner channel, as a channel for, for core CCaaS. Can you maybe illuminate us in, in terms of, is that what it is?
What the partnership strategy is, kind of?
Yeah. The partnership strategy and if they're, if they're, you know, retired, you know, if they can retire quota.
Oh, right.
Things like that, and selling the actual core Five9 product. 'Cause I think we've been over with Barry and Jonathan that these data-led AI solutions are gonna be important going forward. Now we just still try to double-click exactly on what else the partners can bring to Five9.
Yep. Yeah. Just maybe one minute on our, so our partner strategy, we have a strong leader named Jake Butterfield. We call it our balanced route to market strategy. What that means is we're broken up into two categories. Number one, we have the service providers. Think the AT&T, BTs of the world. They bring us leads as well as they're also our service provider routes to market. We have the global GSIs. The Deloittes of the world, they were our partner of the year. We partner with them on a lot of these large enterprise deals. They bring a lot of data into these conversations and they help walk us into opportunities. We have our ISVs. These are some of the AI point solutions like Cresta.
We have a lot of these partners that are calling us up every day and wanna be part of our marketplace, right? They would get access to the Five9 platform through, you know, Voice Stream and Transcript Stream. We have our very strategic partnerships, which are, you know, the CRMs as an example. We are, of the CCaaS partners out there, the only partner that's at summit status with Salesforce. That just means that we have more customers together than anyone else in terms of CCaaS and CRM together. The history of Five9 has always been to really lean in, especially with the CRM players. Some of our competitors have kinda gone back and forth on this, but in the end, it's really a mutual partnership.
When we walk into an opportunity with Salesforce, our sales teams and the Salesforce sales teams are in the same meeting having the same level of conversations. It does retire their quota on the Salesforce side. Each of these partners, like let's take ServiceNow, ServiceNow called our partnership out on their earnings call, this last earnings call. We are getting really deep with them. They have more of a marketplace approach. At the end of the day, all of them are leaning in very heavily with us in terms of wanting to bring us into the conversations. With all the APIs that the product teams are building, we believe we've made it easier to integrate to us from a CRM to CCaaS perspective than our competitors.
We're leaning in on the go-to-market side, making big investments with them and continue to lean in on the technology side.
Were there any specific changes made on the go-to-market side heading into calendar 2025 post the recent sales kickoff?
Yeah. On the ServiceNow side, they opened up kind of some beta integrations. I think we were the first to market with ServiceNow and some of their focuses. On the Salesforce side, you know, obviously there was the Agent Force announcement, right, which created a lot of buzz. We're working on some specific things with them in terms of how we integrate to them. That'll sort of, you know, probably Q3 this year, you'll see more announcements around our partnership with them.
Okay. Thanks. Pause here for a second. Are there any questions from the audience?
For the, just on the AI side, how able are your customers to do it themselves? Like, they obviously have the capability to do it. How much do they think about the monetization? Like, how are you able to do that? Is it monetary? Is it about monetary? Is it, is it something ROI or what it is for you? Like, what's the sort of dollar amount you're able to, the value that it seems to be collecting from you? How much do you think you are collecting?
Yeah. It's.
Right. It's a type of service. 'Cause there's just, as we just said, there's a lot of stuff that it's, it costs money. If you go to the CRM, hey, do this. They say we can take out 5,000 people, we can do 3,000, whatever, whatever the number is.
Yeah.
How does it show up to the, to the customer?
Yeah. First thing, back to the kind of that labor arbitrage ROI that everyone's looking at, right? Hey, if I deploy AI, I can reduce humans. You know, look, there's a lot of, you know, sort of bare thesis out there that humans are going away. We're not seeing that in our customer base, both in terms of the actual metrics. And we track these things just kind of in real time. Customers deploy AI, we can kinda keep a track on their seats. We're seeing like maybe goals of like 5-7% is kind of what's playing out.
Even in the areas where we do have a high ROI use case that deflects a lot of calls or takes calls out with self-service, they deploy those agents into other parts of the business where you need more human empathy. Back to the other question around the actual proving out of the ROI and how do we monetize that. We mentioned earlier that all of our AI products are consumption-based. Most customers in the contact center world still kind of think of seat-based, right? Usually what they do is they do a little bit of math on the consumption and go, hey, where does this generally fit from a seat perspective?
As they deploy it, if the use case is there and we deliver for them with our services team, ultimately they're willing to continue to pay more money, right, for that consumption because the value's there, right? It's kind of a, it's a, it's a good blend. We still have seat-based products and we have the consumption model. At the end of the day, we have customers come to us and, you know, we think a differentiator for us is our flexibility, flexibility on price. The last thing I would say back to, you know, how we sell and position, one of the things that we really recognized in Q2 was customers really asking a lot more questions around the ROI.
We put some, you know, pretty high-powered teams together to put together just changing our motion of how we engage with our customers and bring them the use cases that we know that will deliver high ROI, that we've proven it, we've invested in it. The last piece is our existing customers. A lot of times existing customers, they, we need, we need to help them build the ROI to go make these investments. We've launched something called AI Blueprint, which what we do is we turn on Voice Stream and we listen into our existing customers' conversations. What that does is we then open up AI Insights, which is the new product that the team has built. We can actually identify where are some of the highest, like, hey, 70% of our calls are calling because they're returning this particular product.
They can spot that in real time and then we can go and update their and build a new self-service application. It is kind of in real time continually tuning and training that. We do a lot of the lift, the heavy lifting on the front end, but the application the team has built has been pretty simple to allow customers to continue to, continue to, to change and tune and moderate that.
That particular product, for example, this AI Insights that you turn on to start listening, has been, that particular one has been self-serviced by the majority of the customers that have put that product in because GenAI, GenAI has made this, it is a new product category. It did not, it was not possible to do this product before GenAI and it has made it possible self-service. We are seeing more of that.
Thanks for the call. Like, for the typical customers that are excited to hear AI products up to 12 months, maybe 2 months, maybe even more than that time, what's the sort of financial outlook you're seeing at those times? You know, help understand, how to price in the business. Like, where can you, how do you see something taking this in a vertical or diagonal financial outcome?
Yep.
I can help answer that. A couple quarters ago, we gave a hypothetical example of, you know, if a customer were to achieve 15% automation, then it translates to 30% increase in ARR for us. Now, we actually gave three concrete examples of just that, right? Where the customer actually implemented IVAs and they, in those cases, they actually did not reduce their agent count. It was more of they slowed down the growth. The ARR uplift for us, overall portfolio of that customer was somewhere in the range of, you know, 25-40% plus, right? To your point, it's still AI is, as Barry mentioned earlier, 9% of enterprise subscription revenue, but that grew from 7% a year ago in Q4 2023. We had acceleration in the growth rate of 40% going to 46% quarter to quarter.
While we haven't quantified exactly what the future looks like, it's gonna continue to be the fastest growing part of our product portfolio. While it makes up a smaller portion today, so it doesn't show up as dramatically, it will continue to be a very strong support for our subscription growth going forward.
We're a billion-dollar company. Man, it takes a while to move the needle on energy.
Well said.
Mm-hmm. Maybe as an extension to that, net retention has stabilized recently, Barry, or Bryan. You know, I agree, you know, with, as a percentage of total revenue too. I mean, it's less than 9% in terms of AI, but it is growing so dynamically relative to the, you know, maybe double-digit percentage top-line growth. As you anniversary, it seems like with the million-dollar plus wins, and these are new customers taking on AI. Maybe it might take a little bit more time to the gentleman's question on lapping these deals from an NRR perspective. I guess in multi-part question, can A, would you assume AI could be an expander for NRR sometime in the next 12 months? If not, what are the other levers that you're seeing that could expand that?
Tom, I'm happy to answer that. I'll take it for 2025 and then beyond as well. There are a couple of factors, tailwinds, headwinds. On the headwind side, we already talked about it related to guidance and that impacts the DBRR as well. On the tailwind side, exactly to your point, the AI momentum we expect to continue. That's gonna be, while it's a smaller portion today, it's gonna continue to contribute towards DBRR increase as well. Of course, we have our million-plus ARR customers who represented 56% of subscription revenue in Q4, and they grew 26% year over year. Their DBRR compared to the 108% that we just reported is significantly higher.
When you net all of that out in 2025, we haven't given specifics around where it ends up, but DBRR at the total level will continue to fluctuate in small bands. Going forward, longer term, because of those two tailwinds that are gonna continue, and especially if macro health becomes more healthy, where install base is really poised to accelerate given all this very strong retention that we've had of our customers, we see there a big opportunity in terms of DBRR going higher in the long run.
Okay. Maybe one more check with the audience. Okay. I think we have time for maybe one more question, maybe two.
Jonathan, you mentioned the billion-dollar run rate here. That is a great achievement. Bryan or Barry, back to the financials, just maybe remind us, and again, timing would be very helpful in terms of, you know, what the levers are to expand, you know, gross margins, which are, and you've been very open about this, that these are, these are subscale, and just maybe kind of remind us about the timing and the levers there.
Yeah. I'll start with Q4 first. We had 63.5% gross margin, adjusted gross margin. That was a nice expansion both sequentially and year- over- year. Really three key drivers there. It was revenue growing against fixed and semi-fixed costs. We had the full quarter benefit of the RIF and then just tighter management of expenses across the board. Those will continue into 2025. What we've said is that on an annual basis, gross margin is expected to expand year over- year in 2025. There are two other factors that I want to talk about. Both of them have to do with a mix shift in revenue. Our telecom usage revenue, typically about one to three percentage points of that mix shift toward subscription.
The reason is by design because our larger customers tend to bring their own telephony and our largest partners offer their own telephony. Telecom usage gross margin today is in the 50s versus subscription in the low 70s. That mix shift will naturally help total gross margin. Similarly, on the professional services side, by design, we're enabling our partners to take on more implementations. The revenue on average should grow slower than subscription. Today, professional services gross margin is near break-even, again, shifting toward the 70% plus for subscription. That is why we're confident that the gross margin will continue to increase. Longer term, of course, those same factors and the fact that, you know, we have momentum in AI, which has great gross margins, those are all gonna be helping us to continue expanding margins.
Is there an update on India there by chance, in the margins, the gross margins in that region?
So in,
I was just surprised we didn't need anything.
Yeah. No, absolutely. We've stamped out many locations internationally. We're not done yet, but, you know, we're aggressively expanding our infrastructure, India being one of them. We do expect to start gaining leverage in these locations, but it will be over the long run, right? Yes, that's absolutely going to be a gross margin expander going forward.
I won't try to squeeze one into the next 30 seconds. Thank you guys very much for your time. It's great to hear the story.
All right. Thank you.