Braze, Inc. (BRZE) September 10, 2026 Conference Transcript

Morning, everybody. I'm Tyler Radke, Citi's co-head of U.S. Software. We are joined by Braze co-founder, CEO, Bill Magnuson. Bill, thank you for joining Citi's tech conference. I think the third or fourth year in a row and shortly after earnings.

So I always appreciate it.

You know, there's been a lot of noise in the market and maybe we can just sit on the earnings topic for a moment, just maybe to kick off, could you just walk us through how the quarter played out? You know, I think there were a lot of questions on the dynamics specifically around CRPO growth. So maybe we could hit that just with how you're seeing underlying growth versus some of those metrics.

Yeah, absolutely. So yeah, I mean, overall, great quarter. As we mentioned, bookings was strong within the quarter. And it was also one of the most consistent global results that we've seen probably in the last five years with all of our major regions contributing at a high level. And similarly, trends around our competitive win rates and salesforce productivity all continued into the quarter. And so great to see that. The organic CRPO did take two percentage point down step to 23 in the quarter. And part of that also contributed to by just seasonality of Q4 and Q1 of this year were really big renewals quarters, which is great to get through more of the renewals for the year a little bit earlier on. And so some seasonality contributing there. But overall, we were really happy with the continued bookings momentum from Q4 into Q1 into Q2 and looking at the pipeline and the setup for the second half. So, you know, happy to flow through the beat and, you know, put a raise into the revenue forecast as well for the full year. And we're looking forward to Forge here in a couple of weeks, which historically has always been a big accelerator pipeline for us. And so, you know, I think that our competitive positioning also continues to be in a really strong space, especially vis-a-vis the legacy marketing clouds. And so the continuation of share gain within that part of the, you know, primarily in the enterprise space is something that certainly my sales leaders are looking forward to here through the end of the year and into the next few quarters. And our AI roadmap is really resonating out there in the market. You know, when you look at the pace at which we've been able to bring products to market, the acceleration of the roadmap this year, you know, the fact that Agent Console and Operator got put into people's hands months ahead of schedule. And now, you know, we've done the same with a lot of our work around our remote MCP server, our headless operator, and other work around things like Decisioning Studio Go, all of which you're going to hear a lot more about at Forge. But all of those things have accelerated. And one of the really exciting things about a lot of the Forge updates this year is just how much of it is already in beta or already moving into general availability. This is definitely not a year where there's a bunch of slideware and future promises. The build is all happening in real time. And that means that we're putting capabilities in customers' hands faster. It also means it's driving contributions to, you know, our growth more quickly as well, and is enhancing our competitiveness too, because frankly, in the competitive landscape, I think that we are sprinting and everyone else is still standing still.

Yeah, no, it's great to hear. And I definitely want to hit on some of the AI product stuff, but just a few more kind of financial questions, because they've been topical. How would you characterize just this reclassification of revenue from subscription into ProServe? I think you had talked about maybe that dynamic allowing you to progress through renewals faster. So just how, what happened? How should investors kind of think about that PS versus subscription going forward?

Yeah. So first of all, I'll just take a step back. The motivation for the pricing and packaging change that is now playing out in the numbers, which I'll explain in a bit more detail. But the motivation actually was to do a better job of aligning the professional services delivery with exactly where customers had a distinct willingness to pay for the different types of advisory services that we provide. And you actually, if you go look at our post-sales headcount, you will see that our post-sales headcount is not like swelling at this rate that the professional services revenue, obviously, it appears to be growing. And that's because there, well, there is this kind of accounting shift in terms of how this revenue is labeled. There has not been like a meaning. It's not like there's been a shift in business strategy toward services. In fact, actually, the relative proportion of post-sales headcount is shrinking in the business because we're achieving efficiency gains through operator doing things like delivering technical support and, you know, So one thing that is happening, and I mentioned this on the call, though, is that there are forms of advanced advisory services that we are seeing increased demand for because of some of the some of the increases in like the adversarial environment. Right. It is deliverability is harder than ever. And everyone in this room with an email inbox probably understands why. When you look at the amount of agentic messaging and mail traffic that's out there right now, for brands that actually are trying to play this game the right way, with the right levels of consent and engaging with their loyalty programs and loyal audiences and things like that, simply getting that signal through all the noise is harder than it's ever been. And so we are seeing, you know, incremental demand for things like our premium deliverability services. And that's because your pipes can't be dumped if you're going to actually, you know, drive revenue and get effectiveness out of your messaging and customer engagement strategies. And similarly, on the technical account management side, our TAMs really help customers with the architectural setup of the way that they set up their marketer workflows and the way that they integrate their data flows and things like that. And we're seeing increased demand for that, too, because the AI models that are able to drive enhanced performance for customers need to be fed with the right data and context in order for them to do their job. And there's a ton of enthusiasm for marketers to do things like adopting our MCP server, being able to adopt and run operator, and then additional kind of more bespoke AI workflow work. Like we've got customers that are using Cloud Cowork and then Headless Operator and our MCP servers together where they're, you know, going from kind of full brief to end-to-end strategy. And then actually Headless Operator can go and implement it in Canvas. And so they can actually kind of work in a Cloud Coworkflow, if you will, and then go and see the results of all the programs that got built, like inside the Brace dashboard. And that's really cool, but it requires also, you know, setting up those MCP servers and getting the right secure, like, architectural setup in place and such. And those are places where our technical account managers help people with that. And so we are, you know, a part of this is that we're seeing increased demand for those advisory services. I'll remind everyone that actually the way those advisory services are purchased from Braze is that they're always co-termed with subscriptions and the revenue is recognized ratably. they are all like 90% of those services are recurring and they're tied to the exact same contract length as like everything else. And that's kind of always been the case. And then the other thing, as I mentioned, that has been happening is that we started breaking out certain services that used to be bundled into the platform. And that has caused them from an accounting treatment standpoint to move from being classified under subscription to under professional services. That means that when you do the year-over-year compare, that same revenue was in subscription in the prior period, and now it's in services. But it has traveled, but it is not. And that is causing the relative growth rates to look out of whack. And so if you want to do an apples-to-apples comparison, and we shared this on earnings night as well, but the apples-to-apples percentage of revenue from professional services if we had not done the packaging change would be about four to five percent of the revenue would be in services and so if you want to look at that as a comparison point to just get a better understanding of how the underlying subscription revenue has been growing that you can use that as a number to flow through so got it okay no that that's helpful uh and then on on gross margins you know typically braise has seen some gross margin pressure in Q4 due to holiday messaging.

How should investors think about the gross margin outlet this year?

Yeah, so the seasonality will operate exactly the same as it has in prior years. As a reminder, as well, as a consequence of our revenue being recognized ratably, which means that it's throughout the term or throughout the multi-year terms, in most cases that customers have uh the q4 is a higher activity level order um across across the board it's not you know just sms and whatsapp it's like literally there's also we have more servers online because there's more data flowing and everything else and so um it's just uh it's a quarter with with more activity and therefore uh you know higher infrastructure cogs and so um it will have the same seasonality in gross margin percent um all that being said uh gross margin uh definitely has been a big focus area for us. You know, we were working on it on multiple fronts. The new products that we've been rolling out in the Braze AI family are all priced to be accretive to overall gross margin. Similarly, there are certain workloads that we are able to monetize now for the first time. For instance, like we don't charge for push notifications in Braze. We never have. And we send like, you know, over a trillion of them every year. But as customers start to use our AI capabilities to optimize those push notification campaigns, we'll actually start to make money on those workloads for the first time incrementally. And so that's an exciting forward-looking opportunity as customers start to invoke our more advanced models in more and more of their existing messaging workloads as well. Uh, and, uh, and that, that's, you know, a lot of that is on the foundation too, of, um, the work that we've done to move our whole, um, most, the vast majority at this point of our customer base into our action credits model, um, which, which allows for them to more fungibly utilize those committed credits across different channel types or, um, you know, using, uh, the same credits that they previously would purchase to send an SMS. they can now also use to invoke a reinforcement learning model that's going to enhance the performance, you know, of their upcoming campaigns. We actually, I just got another data point yesterday of a customer testing out a content optimizer in their SMS welcome flow. And they had more than an 80% increase in clicks in the welcome flow. And similarly, I think it was like 81% on clicks and 83% on conversions in terms of uplift from taking the workflow that they were doing before where they were spending you know good amount of money to run a welcome workflow through sms as a channel and they ran a 90 10 test where they put 90 of the traffic on the new one that used content optimizer and they left 10 on the prior strategy as a control and that experiment of course is dead simple to do embrace to be able to go from your old strategy to one that has like an experiment branch on it and the 90 path use content optimizer and they achieved over an 80 increase in the performance of that and so that's just a a great example where there's a whole bunch of of unoptimized workloads that have been previously programmed into canvases and braze that you know we're really looking forward to customers being able to you know go back through like not just build with AI moving into the future, but actually also go back through all the abandoned cart campaigns and the new user onboarding series and the cross-sell messaging and things like that and put more advanced AI into the flows that actually generate that content in order to drive more performance for them. And I think that there's a lot of performance enhancement opportunity for our customers as they go and work through that. And the good news for them from just the burden of doing those builds and modifications is that Operator can now do it for them. We have been rapidly advancing the capabilities of Operator. It now, even just over the last six weeks, we've added the capability for it to independently navigate to all different pages of the dashboard. It can now also build multi-step canvases for customers. And so you can just give it like, hey, I want to build a new welcome series, or I want to build a new abandoned cart campaign. And for instance, like within an abandoned cart campaign, it can actually, it knows now, and we're going to continue to train it on these more advanced strategies, that an abandoned cart campaign should use an agent console step in order to enhance the content personalization that actually gets sent out. And so it's no longer just like, you know, here's the starting trigger and a time delay and then like, you know, send the message. It's also going to include like and send the message based on the intelligent channel for, you know, that given customer and take all the breadcrumbs that you've got along the way around what else the customer maybe looked at recently, the last few products that they were considering between, et cetera, and put all that in a context window. and have the agent actually, you know, use that understanding from the breadcrumbs of the customer journey leading up to that to write a really compelling, you know, abandoned cart, compare and contrast of everything that they were looking at recently or something with, you know, inspiring ideas of a great recipe or a great fitness plan or a great vacation that they might be able to go on, you know, depending on what category or vertical you're operating in, right? And those types of more dynamic, like content optimizing opportunities are now, they're simple to build using operator now. And so I think we're really excited as well, that a lot of the advanced capabilities that frankly, you know, I think Braze has been on the leading edge of building advanced capabilities for our space, you know, for a long time now. And the adoption of those capabilities has kind of always lagged behind what I think the potential of them has been. And that's left a lot of performance on the floor that, frankly, people could have picked up. But for bandwidth and training, there's all these things that kind of got in the way. But now with Operator, it collapses down all the steps to be able to move from us building more advanced capability that can drive uplift to getting built and implemented. you know, tested and deployed.

And in that customer example where they saw this massive increase in click through and conversion rates, like you're, you, you also are benefiting from that, right? Because they're, they're using more credits and the AI capability.

Yeah, absolutely. Absolutely. Cause yeah, the content optimizer that's running as well as the, um, you know, the fact that the, and then of course, when customers, uh, succeed and they engage their customers more than that's more monthly active users, their businesses stronger, right? Like You keep that customer engaged, and so you continue to communicate with them more over the long term. The higher quality and the more effective messaging is, the better our customers grow their businesses, and that obviously results in increased growth for us.

Yeah, yeah, absolutely. Maybe the last financial question, and you're doing a good job with the CFO hat on these, but the beat in the quarter on operating profits was strong. I think some investors were a little underwhelmed by the raise for the full year. So can you just talk about how we're thinking about back half operating income and profitability?

Well, so first of all, we did raise the operating income margin percent from 8 to 8.3 on the guidance for the full year. And so flowed through an increase in profitability for the year, both in absolute and in percentage terms. um what we didn't do is we didn't we didn't flow through the entirety of the qq2 beat into the second half um you know the q3 uh q3 number and consensus frankly was a little out of position didn't take into account forge uh which has been there every year for the last several years and so we'll share responsibility on that one um but uh but yeah i mean we're excited about the investment for forge uh this year forge is going to be bigger and better than ever uh we um you You know, we're in a new venue in Vegas, higher capacity. On the strip too, right? Yep, yep. This year we're, it's a little bit, I don't know, north or south, but we're not right in the heart of it. We're at the Fontainebleau this year. And, you know, we're really excited to bring together that community. I think at this point, Forge is actually the largest focused gathering of marketers, like, for any event in the world. You know, you've got events like Cannes Lions, you know, the advertising festival. But those are kind of overrun by, you know, advertising and large tech companies or what have you. You know, Forge really is a it's a professional retreat for the most ambitious and creative marketers in the world. And we're really excited about, you know, what it means for the business, because it's always a big pipeline accelerator for us. But also it's you know, it's time to build right now. And we've put a lot of really exciting tools into the hands of our community. And if you haven't been to Forge, the entire first day is dedicated to learning and workshops. And we anticipate, you know, doing like in excess of a thousand certifications on the first day of Forge and running a number of hackathon workshops that are going to bring hundreds of people through and introduce them to, you know, our new AI products in person in a lab building hackathon like setting. And so, yeah, really excited about that investment. And then, yeah, and then when you look at the full second half, you know, we held back about a half a million dollars of the Q2 beat in terms of the full year. And the primary use case for that is actually that we're starting our ramping of recruiting for new sellers into next year earlier than we normally do. So that's normally a Q4, kind of start that in Q4 so that they're here in time for kickoff in February. You know, this year we're just starting now, you know, for a few reasons. One is that we've been really happy with the consistency of performance around the world. As I mentioned, you know, we've got within our verticalized teams, as well as within all of our global regions, we've got, you know, strong leadership, strong programming, great momentum across the board. And so there's room to be adding capacity. Sales productivity also has been in a strong position. And then the last reason is also that there's, frankly, a talent war going on out there, right? For people that are adept at selling AI software, there's a lot of competition for those people. And we think that there's a great opportunity for us to start next year with an enhanced sales capacity relative to this year in a meaningful way. But we've got to get those butts in seats. We've got to get them recruited and on ramp. And so we're spending to make sure that we're able to do that as we start next year. But we did all that as well against the backdrop of increasing the full year operating income margin percent, you know, as I mentioned, from 8 to 8.3 percent. And so, you know, I hope that people appreciate that that not only fulfills our prior commitment of 400 basis points year over year, but we're actually now pacing ahead of that. And we're pacing ahead of that even while we're investing more in, you know, building up capacity for next year. And so, you know, from my perspective, like all pointed in the right direction.

Yeah. You brought up a good point earlier where I think historically Braze has built incredibly sophisticated products for marketers, but may be out in some cases out of their technical depth. But AI is making that easier to adopt, right? Because agents can call on this. Like, what are the big bottlenecks and still constraints that you have in terms of customers being able to adopt your AI solutions? I know you talked about some constraints on your side in terms of forward deployed engineers, but just what are those big constraints that's limiting kind of a faster AI rollout?

Yeah, you know, I think it's not just use of AI. I think also when you look at what, for instance, cross-channel strategies are another big priority for us, having customers who are implementing their marketing to take advantage of all the logic that we have to be able to mix and match and fall back to and utilize, you know, different channels tend to not get used as much as they should. Also, if you consider a global business, any strategy that a global business is implementing using, say, SMS or WhatsApp should also be implementing it using Line and Cacao and all of the above. Right. Like if you have an SMS strategy and you are a global or kind of a digital first business, that strategy should also be being implemented using WhatsApp and Line and Kakao so that you can implement, you know, effectively the same strategy in every country around the world. That doesn't happen very frequently, though, because it requires building out, you know, four times as much like content. A lot of it is shared, but, you know, they do have nuanced differences from channel to channel. You also have to have it all set up in advance, right? And people might not have activated all those channels so far, et cetera. And so I think that's a big goal because as operator makes it easier to just implement the right cross-channel strategies, we're going to want to make sure that all of those channels are activated and set up and available to customers so that they can expand the reach of the strategies that they're already running, that they already know are running. And then on the other side, you know, data is the answer, right? Like these models are hungry for context. And that's why we built the Brace data platform in the first place is to make sure that data can flow into Brace quickly, completely, and cheaply, right? And that goal has not changed. And the consumers of it now are voracious. The models would love to have even more signals and even more data. One of the amazing things that we're seeing with Agent Console in particular is actually how good it is at getting signal from sparse data points. We've actually had a number of use cases. We had one that I was talking to a customer about just a few weeks ago where they had an onboarding journey. that they had great they'd optimized the performance of like over time through a lot of experimentation but they had a cohort of of users that they really struggled with performance on and the big reason was that they just didn't have a lot of data about them like these were customers that didn't necessarily come in and like you know immediately navigated to their favorite product category or to their favorite sports teams or whatever right where you provide this really solid signal about what you're interested in. But what they experimented with was just taking the sparse data that they had and putting it into the agent console context window and asking it to try to categorize those customers. And it worked super well. They actually got a really meaningful uplift out of it. And so on the one hand, a big part of the answer to that is data. They're hungry for it um but on the flip side actually uh being able to find uh those like slivers of signal in you know in in a lot of noise uh is one of the things that these models are really really good at too and so um you know it's about getting the right data uh into the system um but you know what the right data is uh is not necessarily like an overwhelming comprehensive amount of it um you know that making sure that the, like, signal-providing context is there and is available in real time, you know, will continue to be a challenge. But that's, like, that's been an important challenge forever. And, you know, obviously, we invest heavily in making sure that customers can get that data into Braze quickly, completely, and cheaply.

Right. With all the progress in AI, both across the industry and the features that you've been releasing at Braze, How has that impacted customer buying cycles at all? How would you sort of characterize the overall environment? It sounded like a pretty broad-based, healthy performance across regions in the quarter. Do you think we're kind of coming out of this area of AI confusion that maybe we saw earlier this year? How would you sort of characterize the overall tone of conversation?

Yeah, I mean, I think overall, you know, Braze is the AI transformation bet for, you know, a broad, broad array of our customers and the productivity gains available by people adopting, like adopting operator and adopting our MCP server are astounding. And so I would say it's involved in and critical in every single deal cycle. Every deal cycle that we do, we're demoing these, you know, marketer productivity and like workflow acceleration enhancements. Similarly, differentiation for enhanced performance is becoming increasingly critical because, and this just goes back to the point I made in the prepared remarks, is that, like, in an environment this adversarial, this difficult, you better not have dumb pipes, Like, you want pipes that are self-learning, self-optimizing, you know, high performance, they've got great observability on them. And that, you know, allows you to then bring a creative vision to life and have it actually get seen by customers and have it actually, you know, drive performance that enhances your revenue growth as a business. And so, you know, those remain the goals for marketers like higher engagement, higher cross-sell, higher, you know, higher LTV, higher subscription retention, et cetera. But they're now able to bring ideas to life substantially faster. One of the interesting things that we're seeing as well that you've seen in the software development world is that as coding sped up, you see this in GitHub's downtime history over the last couple of months as well, that it has led to this explosion in the number of pull requests and therefore also in the demands for doing code review. And we're seeing the same effect happening at Braze, where as operator has sped up the ability for customers to build new canvases and campaigns, the demand for the equivalent of code review, which is all the campaign QA, has skyrocketed. And that's why I mentioned on the call the launch of Egentic Standards, which is us continuing to then also push forward on the whole workflow so that we've sped up the creation side of it enormously. But we now have these systems that are both being built rapidly, which requires more automation for things like QA in the first place. But the other thing is that these systems are increasingly dynamic as well. And so the problem of like the equivalent of QA is like not a static process anymore where you like check all the settings and you say like, yeah, this is good because these systems learn and evolve over time and they interact with each other in various ways, right? You can have, for instance, maybe a program over here that's in the early welcome series is then causing customers to be on a different early journey. And that early journey change might result in the cross-sell that you're trying to do when they've adopted their first product of yours and you're trying to move them to the second product. The nature of optimizing that might change because the welcome journey was more dynamic and adapted. And, you know, similarly, you might see deliverability changes happen over time where, you know, maybe you make a mistake somewhere and you end up with more unsubscribes from email, like, early on. And then you need to go and, like, save that customer cohort later on through SMS as, like, a channel and get them to, like, reactivate. But that also has higher budget needs, right, because those cost more. And so great example where, like, you might have approved both of those campaigns in isolation, but then the way they interact with each other over time is going to lead to performance changes and, like, budget changes, et cetera. And you need the ability to, you know, monitor and observe that, too. And so we're thinking a lot about not – and by the way, the answer to that right now is actually utilizing our MCP servers to be able to pull in the results and be able to automatically analyze that. And, you know, you've got customers who are using things like Slack bot AI in their channels with their marketing teams or, you know, like cloud tag living inside of your Slack channel where where your team hangs out and being able to analyze results and pull in insights like that so that people can kind of pay closer attention to how the system is evolving and adapting over time. And that's an exciting new world. You know, we're looking at a capability that is not just for the single player marketer, you know, trying to like help them write copy faster, but actually having the entire system be able to be built out and dynamically monitored and QA, you know, being something that's not just a static concept, but actually being more like the wardens that, you know, live within the system and like, you know, keep an eye on how they are dynamically evolving over time. And all of that is not to mention also like, you know, instances of fraud are going up out there too. And things like traffic pumping attacks and all sorts of other anomalous behavior that might show up as various groups or different AI systems now are like attacking your first party infrastructure, your loyalty programs or like whatever it is. And being able to have strong defenses and vigilance against activity like that is increasingly important, too. And so, you know, those are also areas where our technical account managers are helping people with the secure design of a lot of their, like, marketing infrastructure. And, you know, it's a difficult adversarial world out there right now, too. And so we're making sure that we're bringing AI innovation to all sides of the problem.

Yeah, I'm sure it's topical with the MetaMuse and some of these consumer-facing chatbots that have exploded in popularity over the last few weeks. Yeah, for sure. I guess just in closing, a lot of exciting things happening. It's kind of the accelerated go-to-market investments, the AI product innovation, which we'll talk about at Forge. If you're sitting here next year, where do you want this company to be? Where are you aspiring to take it if you had to break it down into, you know, two or three high-level goals?

Yeah, I mean, I think we are in a golden era of product advancement at Braze right now. The foundations that we have built over the last 15 years with our vertical integration, the Braze data platform, and, you know, the full stream processor and Canvas as a programming environment have come together in a way that, you know, we already had most of the pieces assembled for a full agentic harness for, you know, marketing technology effectively. And we've only seen that accelerate with the innovation that we've been driving through operator and man watching operator interface with Canvas as a programming environment is like magic in action. And, and, and we, you know, similarly have really exciting research going on across a number of different AI technologies to be able to drive these advanced models, to be able to do automatic optimization of content and decision making across a number of different price points and different levels of sophistication. And so we're focused on building out an entire family of advanced AI. But I think that it is the, you mentioned it up front, it's the first time in our history, I think where we are simultaneously advancing the power and capability of Braze and making it easier to use and faster to use at the same time, you know, that advances have always in the past come with a higher education burden and more complexity and harder to build and everything else. and like man it's it's like it's only upside now uh it's it's really great to see um you know and i think that how advanced braise has is uh has in the past been kind of both a blessing and a curse and nowadays it's only a blessing it's just great to see so awesome great well let's uh leave it

there i think we're out of time thank you everyone for joining bill appreciate you joining us shortly after earnings and look forward to seeing you out in vegas yeah a couple weeks absolutely thanks for