Manhattan Associates, Inc. (MANH) Earnings Call Transcript & Summary

Perfect. Let's go ahead and get started here. So thank you, everyone, for joining. My name is Chris Quintero. I'm the back-office software analyst here at Morgan Stanley. And I'm really excited to be joined by the Manhattan team here, Eric Clark, CEO; and Sanjeev Siotia, CTO. Thank you all for being here.

Thank you. Thanks for having us.

Absolutely. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative.

So to kick things off, for investors who maybe aren't as familiar with Manhattan Associates, maybe you can give us a quick overview of what you all do, key products, who your core customers are.

Yes. So Manhattan has been in business for 30-plus years, always focused on the supply chain and commerce space. So our key products are across warehouse management, transportation management, order management, point-of-sale and supply chain planning. And about 10, 12 years ago, Manhattan recreated itself as a cloud company. So now all of those products are available on our cloud platform that we call the active platform. And over the past 10 years, we've been in the process of converting our on-prem customers into the cloud. At the same time, we've been aggressively in the market going after new logo customers and taking market share from the competition.

Got it. That's a helpful overview. Let's just jump right into AI. Eric, Sanjeev, you guys both have put out some really interesting blog posts as it relates to what the future of software looks like in an agentic world. Maybe give us a quick recap of kind of what you all shared in those blog posts.

Sure. Maybe I'll kick off and let you go deeper, Sanjeev. But yes, clearly, there's been a lot of talk about AI and how that's going to impact software and SaaS. We're very clear believers that AI is going to benefit and make some software more valuable. And there are other software players that likely will be hurt by that. And Sanjeev will share more about kind of the differences. But one of the things that makes us unique in this conversation is the platform that Sanjeev and team created a dozen years ago is all microservices, all API-driven. So when we go to our customers and our prospects and talk about AI, we don't have to start the conversation talking about pulling out all your data and moving it to a data lake and indexing this and data security here and duplication and latency and all those things. All of our AI that we deliver is native in the platform. So we're using the same APIs that a user could use to run the software. So that's a very different conversation about AI. And when we talk to our customers, we can have them live up and running on an AI demo today. Because, again, we are versionless software. All of our customers already have access to our AI agents. They just have to subscribe to them. So that puts us in a very different position, a very different part of the conversation than a lot of this debate that's going on out there about how AI is going to impact software.

Yes. I'll add a little bit to just how the investors are taking the whole narrative around cost and AI is going to replace software because cost is going to drop, right? I mean it's not paper towel. It's not a commodity that you buy. No enterprise CEO is going to kind of throw out their system because something else is 10% or 20% cheaper, right? I mean iPhone didn't disrupt BlackBerry because iPhone was cheaper, right? Digital cameras didn't replace those things because they were cheaper. So most of the time when you see disruptions, they come from because whatever the next thing is, is better, right? So I think the right discussion that we had is how can you innovate faster? What can you innovate, right? That was kind of part of the article. And I'm not even go to the math, right? The 10%, 20%, even if I concede that something will become 20% cheaper, that's not good enough to kind of go replace these complex enterprise systems, right? It takes a lot to build enterprise software and coding is just a very small piece of it. So I kind of use the word 6% policy is if you do the math, I mean, you're looking at 6%, 7% of the total cost. So how much -- you cannot kind of make software cheaper, right? So I think the focus really needs to shift in is how will AI disrupt is how can you become more innovative? Can you deliver with velocity? Can you do a lot more with it, right? That's the discussion that we had. And I would be happy if that discussion started as this narrative around innovation versus, hey, it's going to be cheaper, so let's just write out software.

Yes. Now we'll definitely talk about the innovation piece. But maybe just on the 6%, give people a quick overview how you get to that number and what it exactly is?

So roughly, if you kind of look at any organization, most public companies, you can look at the data, they spend somewhere between 10% to 25% on R&D, right? And if you look at the R&D cost, 25% of the time is essentially spent on writing code actually. A lot of time is spent on understanding requirements, business requirements, figuring out how do you architect it, how do you scale it, how do you kind of you put in production, how do you support it. That's where a lot of cost goes in. And people are focused on this narrow bucket of just the software piece. If I even take the high end of 25%, multiply by 25% of the software, you get to 6.25%. And that's basically what everybody is kind of writing out software because maybe the 6% will become cheaper.

Yes. No, that aligns pretty closely with what our team, Sanjit Singh, our team covers the infrastructure side of software, and he's always said that the actual coding part of building software is a minority part of what a developer's time is actually spent on. Let's talk about the innovation, the opportunity Manhattan has with AI. How do you all view as your kind of right to win with AI? And what are some of the early capabilities? I know you've launched some agents and agent building platform, too. So maybe hit on that as well.

Yes. So I'll kind of pick on -- Eric kind of touched upon a few of those things already, right? So if you look at us, the unique approach we have taken it goes back to our build 12 years back. is we are very API-centric. Everything is API first, right? So the way I describe agentic AI in a very simple way is you give it an objective and you run a think-see-do loop, right? You think about what the problem statement is, break it down, you look at the data and then you take actions, right, until you get to the outcome. And that's when the think-see-do loop can finish. The think part of it for all of us will come from the hyperscalers, right? The big guys, they're building the models, none of us are building the models. And that will be a commodity almost to everybody accessible to anybody. The see-and-do is where we feel our opportunity and differentiation comes from is the see part comes from -- for us comes from our APIs. For everybody else, like Eric mentioned, they have to create a data lake. So the conversation to even start with an AI starts with, okay, let's get all your data into this data lake. Now let's figure out who has access to this data, let's put governance around it, and then you can start figuring out who can use AI, right? With us, that question doesn't even come because our APIs are already just naturally secure. When you get to the do part of it, it becomes even more because every action, the whole system is accessible to AI to actually take actions. rest of them have to figure out what APIs to kind of even put for AI to go execute. So I think the whole -- the think-see-do loop, we can put in people's production today, right, versus somebody else who will take 3, 5, 6 months to even figure out how to go execute that. I think that gives us a very big advantage.

Yes. Maybe, Sanjeev, sticking with you, you've been one of the key architects of building that micro services active platform Manhattan in the cloud. Curious what parallels you can draw, if any, between kind of the AI evolution we're seeing now with that kind of initial cloud evolution?

Yes. So actually, funny, what I would say is our cloud journey was our AI journey is the AI evolution, and I'll kind of explain this in a second, right? So when we started building for the cloud, one of the things we kind of made a decision was we'll build everything API first. Why did we decide that? We have seen historically that every 10, 15 years, user interfaces changed, right, from green screens to the GUI to web screen to mobile phone. And when we were building in 2014, we wanted to future-proof ourselves. So we said, okay, we do not know what the user interface will be 5 years, 10 years from now. So let's kind of just make sure that we kind of build everything headless so we can kind of live in a world on whatever the new user interface will be. So that's kind of part 1. Part 2 was composability, right? When we built this whole thing, we wanted to make sure that we can compose our solutions, right? So we kind of broke traditional boundaries of WMS, TMS, everything else. We thought of it as an overall holistic solution and build these things as composable pieces, which can be stitched together, right? Still AI came through, people were kind of writing those workflows on their own. Now AI can help those workflows. And the third big piece, which we have not talked too much about in the past, and I've started kind of talking a lot about it because this whole code generation has started coming is when you build enterprise systems, it takes time. It takes anywhere from 3 to 5 years to kind of put something really out there. And what happens in those cases is the stuff you build in year 1 looks very different than the stuff you build in year 5 because -- and you build a tech debt in not all systems. So what we did in our approach from a cloud journey was we took actually 6 to 8 months in the initial part of it to write a generator, where people can express -- the developers can express their business intent and we can actually generate code. So we actually -- today, we have about roughly somewhere between 60 million to 65 million lines of code. 75% of that is generated, right? We do not write that piece of code, that's all completely generated. Now the advantage of that was not efficiency, right? We didn't do it because I mean, efficiency was part of it. But the big advantage was how do you kind of keep this tech debt out, right? How do you kind of make sure that when you have a new requirement, you can put it in the core and that goes across the product. So that was the core. Now that became our AI evolution because all 3 pieces which I talked about, right, the user interface, AI can be that user interface. You can kind of put the head of the AI to kind of really make that UX very conversational. The composability, AI needs that composability so you can kind of take the right pieces and build those workflows together, right? And then this code generation. AI is built into our code now, right? So there are a lot of use cases of AI, like we write a lot of natural language audits. We can kind of figure out when you're talking to us in a different language, we can even interpret the synonyms out of it using AI. And that's built into the code generated every night, right? So it's very deterministic. So I think our cloud journey was our AI evolution, you can call it serendipity, architectural foresight, whatever you want to call it, but that is the same journey. So we are -- we have been on this AI native journey, I will say, for 10 years now.

Yes. Yes. Let's talk about some of the newer stuff you all have launched, the agentic AI that went commercial in January of this year. You've introduced agent foundry, active agents. What are some of those? And what are some of the key use cases you're really targeting with some of the agents and the customers that are building on foundry, what are they building?

Yes. So as we kind of think through these AIs, we look at it into 4 buckets of things. The first bucket is UX because we believe that AI will fundamentally change how all of us interact with any kind of software. So there will be a pretty big change from that. So that's the UX bucket. The second piece as any complex software provider in the ISV is how do you accelerate time to value. So what can you do using AI to accelerate time to value. The third piece is productivity. Every is looking for how can I make something more productive, impact the top line, bottom line, FTEs. And the fourth piece is data insights. How can I draw more data insights using AI from my data. So we're looking at those 4 buckets and every agent we have put out there, they fit into one of these buckets, right? Sometimes an agent can fit into multiple of these buckets, but that's been the core, right? I'll give you -- and you asked about -- and then agent foundry, right, because we are completely AI -- API native, we exposed our agent foundry so people can write their own things in do loop, right? They have their use cases. They can put it together very quickly, and they can get benefit from it, right? We've talked about -- publicly about one thing about Eaton, so I'm going to use that example. At Eaton, RFDs created an agent in about a week, 1.5 weeks. They had a problem with around their dock cleanups, and it was an exception process and somebody would have to kind of sit down every 30 minutes, check something, see if that exceptions happen and then take an action on it, right? Our FTE team was able to quickly create an agent and take that problem away. And the biggest impact to that was not just the saving of the FTE, they increased their fill rate by 3%, right, in that warehouse. That's a very significant impact to the bottom line. So that single agent provided enough value to them to justify any kind of cost we would ask them to pay.

So when we go to market with our AI solutions, and you mentioned the foundry and agents, et cetera, what we sell is access to a standard set of base agents across every product. So things like a wave agent, a labor planning agent, et cetera, and the ability to modify any of those agents because many of our customers create extensions on our platform. So they're not all using our software the exact same way. So they might want to modify those agents to make them more valuable to them. And then finally, we also give them access to the foundry that Sanjeev talked about, where they can build an agent from scratch, right, just handle a custom challenge that only they have or maybe to address, again, extensions that they've built because their process may be very unique. So that whole landscape of agentic AI is available through one SKU that we offer, and then they have access to use that as much as they need to.

Got it. I want to go back to kind of the whole kind of cloud transition and the opportunity there. Supply chain software as an entire kind of like subsector within the software has been a bit slower to move to the cloud. I think it's got the lowest percentage on the cloud, around 40% is what we estimate. Curious why do you think that is? And how do you think about the evolution of how that continues to progress into the cloud over time?

I think a big part of that is, number one, supply chain is mission-critical software, right? It is the backbone of what they're doing. And number two, it's very complex. And a lot of these customers that haven't moved to the cloud yet, they have memory of -- for their existence, every 5 to 10 years, they do a software upgrade. And the old on-prem model, right, when you get ready to do that upgrade, it's like open heart surgery. And it is a massive disruption. These projects often went long, went over budget, and they've got pain and scars from those things. So I think there are customers that are -- they try to sweat that project as long as they can and get value out of that software as long as they can before they go through the next one. So one of the things that we're doing with our customer base is talking with all of our on-prem customers and helping them realize that memory from the past is not what you're going to experience this time. And we've built automation and leveraged AI to simplify that transition from our on-prem products to our cloud products. And to go even further, midway through last year, we introduced fixed price and fixed time line for those conversion opportunities. So we're taking the risk out of it for them and essentially just making the carrot bigger. And we're making it very clear, if you're running a 2020 version of software, you haven't had new updates in quite some time. So let's look at all the features and functions that have been made available since then. And on top of that, now we've got agentic AI available on the platform. So the gap between what those on-prem customers are running and what they could be running in the cloud keeps getting bigger and bigger. And then I'll go back to where I started. This is mission-critical software. It is strategic. If they're trying to be strategic and differentiated on old software, they're not going to have success there very long.

Yes. You mentioned a little bit about that fixed fee implementation. What are some of the other kind of incentives? Obviously, the agents are only available in the cloud, too, but what are some of the other incentives you've rolled out to help migrate? You still have about 70% to 80% of the customer base that's still on the on-premise products.

Yes. And if you think about it, we rolled out our cloud version of warehouse management just over 5 years ago. So we've got a lot of customers that are still in that window of -- they've put in their products 5 years ago, and they haven't really taken the next step to change. So I think naturally, that will start to pick up and accelerate over the next year or 2, but we're doing a lot to help accelerate that. So as we talked about the fixed fee and the fixed time line, making it really clear the difference between what they're running today and what they could be running if they move to the cloud and then going back every quarter and showing them. Once you do this move to the cloud, you'll never have another conversion again, right? It's versionless software. You get new updates every quarter. So we keep showing them every quarter. If you were on the cloud, this is what you would have gotten new this quarter. So I think we're creating the carrot out there. And without a doubt, our pipeline and conversations around conversions today are vastly different than the volume we had a year ago. We made a lot of progress in being more proactive and consultative in helping those customers understand when and why it makes sense for them to convert to the cloud.

Yes. And investors are rightly super excited about that whole cloud conversion opportunity for you all. But when I -- when I've spoken to some of your customers, one of the reasons they've given me why they haven't really wanted to move to the cloud has been we've done a ton of customizations on the on-premise product, kind of what you were saying earlier. Curious kind of how you think about getting past that to help migrate those customers? Is that really where kind of the agent foundry comes in to help them build those customizations extensions on the platform?

Yes. So some of it, right, if you think about -- and we have done a lot of conversions, right? We've still done 22%. We have converted already. So we have a lot of experience on this thing. So when we look at conversion customers, what we find is the new product just base features, it covers 30%, 40% of the extensions, which they've already built, right? So those extensions just completely go away. The remaining part of the extensions, all of these customers were used to this 5-year or 10-year cycle where they'll pay us the onetime upgrade fees. And with the help of AI, with help of every other tools we have, we can get them converted into this in a little bit less cost than they were paying for natural upgrades, right? So those 2 factors combined, that takes that objection away. And some people are just a little bit more risk averse. And for those risk covers, we have enough proof points now over the last 3 years on the others who have converted, right? So I think that will give them another incentive and kind of take away their -- the questioning of the -- can you move to the new version without any issues.

Yes, absolutely. Let's switch over to go-to-market. Eric, that's been a big focus of yours. You made a lot of changes there. So can you give us a quick overview of what have some of those changes been and how they progressed versus your initial set of expectations?

Yes, definitely. So when I joined Manhattan about a year ago, one of the statements that I made was I think we are often underappreciated in the market and particularly outside of WMS. When it comes to warehouse, I don't think a deal happens in the market that we're not aware of. We've been in it long enough. We've been ready to leader for 20 years or something, right? So we get invited to every deal. But across those other products that I mentioned, that's not always true. So we wanted to invest in sales and marketing and make sure that we got invited to all of those opportunities. So one of the things that we did is we took Bob Powell, who had been running sales in the U.S. for quite some time. He's a 20-year veteran of the company. We made him our Global Chief Sales Officer. And that enabled us to start taking some of the practices that we're having success with in the U.S. and globalize them. So that included things like building out product specialist teams. So -- we have a lot of veterans like Bob that have been here 15, 20, 25 years. And you can assume that most of those people, the bulk of their experience is warehouse. And the -- when you go to try to sell point of sale, you're competing with different people, right? You're competing with different software players. You're selling to different people in the company. It's a completely different sale than warehouse. So we recognized that we needed to build product specialist sales teams for point-of-sale and transportation and supply chain planning, et cetera. So we started building those out midyear last year. Then we also hired Greg Betz from Microsoft to come in as our COO to build some programs around conversions and programs around renewals to make sure that we were maximizing the renewal opportunity, not just the price uplift, but also the cross-sell and upsell opportunity across every one of those renewals. So -- and then you add on top of all of that, this AI opportunity. First time since we've -- in the history of our company that we've had an opportunity to go sell something to every single one of our cloud customers that there's no deployment cycle. There's no ramp. We turn it on, and it's an uplift to the revenue. So that kind of near-term revenue growth has not existed for us in the past. So -- and then on top of all that, we brought in a new Chief Marketing Officer, who's really helping us in terms of market awareness and brand awareness outside of the warehouse management space. So I think where we sit today, the amount of opportunities in the pipeline, the ability to execute is vastly different than where we were a year ago. And I'll add one more to that, that's kind of related to this, but not exactly in the sales space. We've turned our services team into a sales engine as well. And our services team traditionally has been follow the sales team and deploy product after it gets sold. But our services team is the primary engine around conversions. They have built and support so many of these on-prem customers. They know what it takes to convert them. And they've also gone out and looked at customers where maybe our customer has subscribed to 50 warehouses, but they've only deployed 30. So going out and having conversations with them about fixed fee deployment of the next 20. Look, I know what it takes to do this. I know what it looks like. I know where it's going. Let me convert the rest of your warehouses, accelerate their ROI and accelerate our revenue growth at the same time.

Yes. So a lot of it is getting yourself into more of these deals that you weren't previously, expanding that brand recognition, bring your best-in-class products to those deals?

Exactly. It's volume of deals, and that's volume of deals that we're competing for in the market, but also just volume of services deals. So there's a big opportunity -- a big services opportunity within our installed base to go deploy products faster. I always tell the team, there's 2 ways we can grow our cloud revenue faster. One is sell more. Two is deploy faster because we sell based on these prescribed ramps. But if we can deliver faster than that and ramp faster than that, cloud revenue grows faster. So we've got the services team focused on that now.

Absolutely. partnerships. That's also been a key part of your strategy, too, system integrators, the big ones, but also some ISVs, too, right? Google Marketplace, Shopify, I think Google Marketplace was the biggest deal in Q2 that was influenced by that partnership. So talk to us a bit how you've evolved the partnership strategy, especially in relation to your services business.

Yes. Yes. So then that's another thing we've got Greg Betz focused on our COO. So driving partnerships, technology partnerships as well as the big GSI partnerships. So I'll start with technology. We announced last year at our Momentum conference in May that we've always been a big partner with Google, but we entered the Google Cloud Marketplace. That's been a driver of new opportunities and new logos. You mentioned our biggest deal in Q2. I think if you look at all of the new logo bookings we did last year, more than 10% went through the marketplace. So it's significant, and we see growing pipeline in that area as well. Shopify has been big for us in the order management space, where we have some good synergies with them. We're working on some additional software and technology partners that we'll be announcing at momentum this year. But we've also just announced a big change in our partner program with GSIs and our services partners that are rolling out software with us. And we've got a lot of those partners that are really excited about what we announced at our sales kickoff in January. So a whole lot more clarity around what we expect of them, what they can expect from us. We've given them access to demo environments, given them access to more of our training and education, our proactive tool set, our AI tool sets so they can invest and build industry-specific templates that then they can take to market and advocate on our behalf. And again, just using those partners to be more feet on the street and get involved in more of those deals across all of the other products outside of WMS.

Yes. Is also the strategy to kind of capture some of those like SAP, on-prem to cloud migrations as well. We've coined this term ERP super cycle. As we get closer to that 2027 deadline, are these GSI partnerships really going to be helpful to drive some of those new customer additions?

They're absolutely super important in that area because I think whether you're talking about POS or one of these other products where maybe there are deals in the market that we haven't been invited to, I promise you, at least one of our partners is aware of those deals. So we need them to bring those deals to us. And the same is true in this SAP cycle of moving from on-prem to the cloud, at least one of our partners is aware of all of those SAP customers that are moving. And many of those customers, as they have to make this change and move to the cloud, they want to look at what's available, what are the best-of-breed products. And when they compete, when they put supply chain out there, whether it's warehouse or transportation, order management, et cetera, when they put it out there for competition, we win. And our win rates show that. We continue to have 70-plus percent win rates against -- across all of our product sets. So again, more at bats, more opportunity to compete, and we feel good about our chances of winning.

Yes. Maybe give us a quick overview of like who some of those key competitors are and where do you really differentiate versus those?

Yes. So that's one of the things that we've gotten better about this year. When you asked that question of who the key competitors are, I think the habit was to name the ones that were the key competitors in the warehouse space, right? And those are the ERPs and the Blue Yonders and people like that. But it's very different in the POS, and it's different than supply chain planning. So we've gotten much more specific about who our competitors are across each of our products and make sure that we've got the marketing, the sales and the go-to-market to address each one of those. But again, the positive side is across all of those products, 70-plus percent win rate.

Yes. Let's switch over to services. That piece revenue returned to growth in Q4. You're guiding to about 3% growth in '26. What's driving that recovery? And how sustainable do you think that can be on the services side?

Yes. So we guided to the revenue growth for this year. We also said in the January earnings call that at that point, we had already added about 100 services headcount this year. You compare that to a year ago where we actually removed about 100 services headcount. So we're in a very different position in terms of demand. And you should assume that every project that we're doing in services now, we're doing it faster than we were doing it a year ago. And we continue to invest in that speed and acceleration. So the reason that services is growing is because we're doing a whole lot more projects. And that healthy really strong new logo that we did last year and particularly in Q4, I think where 55% of our bookings last year were new logos, 75% in Q4 was new logo. So that set us up in a really good position. But also think about all of this focus on conversion and fixed price conversion and all the focus on expansion at the time of renewal and then the focus on our services team driving how do we go deploy everything that's already sold, there's just more deals. There are more services deals out there. So -- we track go-lives by quarter for each one of our products, and we just continue to set high bar marks in terms of go-lives every quarter. So that's a long answer. The short answer is volume. There's just a whole lot of volume and a whole lot of interest. And then the last one on top of that, Sanjeev mentioned the forward deployed engineers around AI. I think for a long time, people have talked about us as a software company, we have a large services team. Well, our services team, we hire from the best engineering schools in the country. They are engineers. So nobody is better prepared than Manhattan for this AI push, right? We're able to spin up forward deployed engineering teams so that when we sell these proof of concepts, not only -- we're not selling it to them and saying, good luck. We're helping them deploy it. We're helping them learn how to build custom agents, helping them learn how to modify agents, making sure that they find value in the AI that we deliver.

Yes, really being a close partner to those customers. You mentioned that a little bit, but the renewal opportunity. As you mentioned, cloud product first went live in 2020. You've added a bunch of new ones since then. And as those customers start to come up for renewal, you cross-sell into those existing customers. So talk to us about what's the magnitude of the opportunity? And what's the timing of when you expect those large renewals to happen?

Yes. So you can, like you said, kind of do the math on when we launched our different cloud products and make some assumptions if our average deal length is about 5 years of when those renewals are happening. Now that's not exact science because there are different length of deals, but it's no secret that in 2026, we are going to have a big opportunity around renewals. And as I mentioned, halfway through last year, we set up this dedicated renewals team to help maximize that opportunity. What I mean by that is in the old account manager program, maybe an account manager did a renewal every 1 or 2 years on one of their accounts. So you don't take a whole lot of learnings from one to the next. And now with this dedicated team that's working with the account managers, we've got a going-in position of what's the price uplift going to be, what's the duration going to be? What are the logical upsells and cross-sells based on what they're running today. And we set ourselves up for a whole lot more success in maximizing that opportunity. And then the second thing that comes with that is, I mentioned we start with a price uplift and a duration. We like to start the conversation around renewal at 3 years because I'd like to have this conversation again in 3 years. I don't want to wait 5 years or 7 years. So one of the biggest pushbacks from customers is they want a longer duration on the renewal. And of course, it's a give and take. Let's talk about what else you want to buy, what you want to add on and really making that process smooth and in partnership to find the value that they're looking for and the value that we're looking for, that's what we're talking about in maximizing the renewal cycle.

Got it. Let's talk about RPO. That reached $2.2 billion in Q4, up 25%. I think you mentioned record cloud bookings. You've guided to about 18%, 20% growth for that metric in '26. So could you walk us through what are some of those building blocks to get you to that 2026 growth rate?

Yes. So -- and let me tie it back to a little bit to this renewal cycle. I mentioned -- I kind of like the idea of a 3-year renewal and get another bite at the apple in 3 years. But if you think about it, if I do a 3-year renewal versus a 7-year renewal, that has a pretty significant difference in how much RPO we book, but it actually has no impact at all on our growth rate in the next 3 years. And that's why we started -- in our last earnings call, we shared the ramped ARR number because I want to make it more clear what's really happening in that ARR growth. And I think historically, people have looked at RPO as the indicator of ARR growth. And as renewals become a bigger and bigger part of our bookings, RPO is going to be a less clear indicator of ARR growth. So that's the reason for doing that.

Got it. Longer term, you guys have talked about sustaining that 20% plus kind of cloud subscription growth rate for multiple years. What ultimately gives you the confidence in that type of durability for that metric?

Yes. I think we've always said one of the great things about our model is it's very, very predictable. I talked about when we sell a deal, there's a ramp and a clear view of how that revenue is going to ramp. So that gives us great confidence and great revenue visibility growth for the next several years. The ramped ARR that we shared is a 4-year ARR of 23% growth year-over-year, right? So all of the indicators are there that gives us strong confidence in what we need to do from a new bookings perspective to continue that 20-plus percent growth rate and not only continue that 20-plus percent growth rate, but we're really focused on reaccelerating and growing that to a bigger and bigger number in the future.

Yes, absolutely. Let's shift to margins before I open it up to the audience here. Margins are coming down a little bit in 2026. Your maintenance and license revenue, that attrition is accelerating a bit more as you guys have talked about. How should investors think about that dynamic? Is this just a kind of onetime thing? Is this something that's going to continue for a couple of years? And how do we think about getting back to that margin expansion cadence?

Yes. So for a guidance on margin, we kind of use the same methodology that we've used in the past couple of years, which is remove the maintenance and license that is declining as we shift people from on-prem to the cloud. But when you remove that, we're 75 basis points improvement year-over-year. So that's the same thing we've guided to the past couple of years. We guided there again this year. If you look at the history, we've got a history of beating that. And our goal is that we would continue to underpromise and overdeliver. But you'll also see that what we are forecasting in terms of that drop in maintenance and license is bigger than what we actually did last year. Reason for that is we're expecting to see more and more traction in this conversion cycle.

Any questions from the audience here? I must be asking all the right questions.

Yes, exactly.

Cool. Maybe last one here, Eric, for me. And Sanjeev will love your perspective, too. As you both reflect on 2025, what are some of the key lessons that you learned that you're going to apply for 2026 in terms of things that worked really well as well as things that you're looking to improve upon?

Yes. So when I look back at 2025, I think one of the big highlights is record bookings performance and 55% of our bookings was new logo. Historically, we've always said that the bookings will come in one-thirds, 1/3 in new logo, 1/3 in conversion and 1/3 in expansion. And I've kind of guided that in '26, we'll get back to 1/3, but I want to be really clear, we're not going to do that by new logos coming down to meet the other 2. We're going to do that by all this focus we have on conversion and expansion so that we can operate at the same level that we've been operating at new logo. So I think that's the big aha moment and those are the investments that we made last year to really drive those other 2 pieces up to and perform at the level that new logo is.

And I think what I'll add is I think we've done a very good job over the last 10, 12 years innovating. We don't tell our story about -- we don't talk about it well. We don't communicate it very well. So I think for me, for 2026, the big part of it becomes is how do we start telling our story a lot more, which is why I saw the blog post for me and probably see some more coming out of it. You just start telling our story and let think people know what we really actually do.

So expect more blog posts?

Yes.

Absolutely. Awesome. We can end it there. Eric, Sanjeev, thank you so much.

Thank you. Appreciate it.