Elastic N.V. (ESTC) Earnings Call Transcript & Summary

Welcome to Citi's tech conference. I'm Tyler Radke Co-Head of U.S. Software. Thanks for bearing with us with the elevators and stairs. I know I got my exercise going up and down the stairs. So we're happy to have Elastic to kick things off here in the afternoon. We have the CFO and COO, Janesh Moorjani in the house with us. And before we get into the recent results, which I know are very topical, there's a lot of investors in here that may have different familiarity with Elastic as the company. Can you just quickly give a synopsis of who Elastic is? And how does that vision look with all the excitement around AI?

Yes, I'm happy to tell it and thanks for having us here today. It is great to be here again in a new venue and everything. So thank you for that. Elastic at its core is a search AI company and was founded back in 2012 on this premise that the power of search technology can be applied to many, many different use cases. And so any time you have very large pools of particularly messy unstructured data, the bigger those pools of data, the better we shine in some senses. And any time you are you're looking for meaningful results or meaningful information within those large pool of data, that's a great use case for Elastic. So you can think of us at our core as a search engine, and we get applied to many different used cases. It search in its most traditional form the way you think about it, whether that's search power on a website or a shopping cart experience or any other way you typically think of search as well as for use cases like security and particularly within security, the SIEM use case, which is fundamentally a data problem or an observability for logging, log analytics, which is again, and fundamentally a data problem where you're trying to analyze these large pools of messy unstructured data. And over the last few years, as Gen AI has become more popular, that's actually been a fairly good catalyst for us in terms of thinking about the way people think about search and the ability to use search to solve all kinds of business problems.

Got it. Great. So let's dive into it. Your fresh of Q1 results last week, which I think in terms of how the quarter played out from a revenue consumption perspective, I think things look relatively straightforward but you had an unforeseen go-to-market impact that led you to have a shortfall on commitments and ultimately, you cut the full year guide by a couple of points. Take us through what happened. What was kind of the biggest surprise that you saw during the quarter?

Yes. I think as you mentioned, Tyler, looking back at sort of the history for the quarter, all of the metrics like the reported revenue for the quarter, our cloud revenue growth, our customer statistics, count of Gen AI customers, including in the larger pool of more than $100,000 of ACV, the customer count in that category, I think all of those things played out really nicely for us, including the net expansion rate, which ticked up slightly to 112%. So all of those metrics played out really nicely for us. I think where we had the biggest issue and the reason it was a dissatisfying quarter from our perspective was we came up short in terms of new sales, essentially customer commitments or what the sales team sells to customers. And the main reason for that was certain segmentation changes that we made at the start of the year entering the new year. Think most people are familiar that every year, most software companies in the industry will make changes to coverage models, territories and segmentation. We were no stranger to that in the past, we would make certain changes as well. Over the last several years, as we've continued to increase our focus on selling within the enterprise and continuing to move further up the stack, if you will, we've continued to make changes and modified our go-to-market approach slightly over the years. There was no fundamental shift in terms of our overall strategy from a go-to-market standpoint. We continue to think of within the United States, we continue to think of our accounts as segmented as strategic accounts, enterprise accounts and then within commercial, you've got mid-market and general business, but broadly thinking of that as commercial. This was also unrelated to the SMB change that we made a couple of years ago. But what we did was three big changes as we entered this fiscal year. One was in the strategic account segment. We added some additional accounts there, but -- and also added more reps, but the net result of that, we actually have fewer accounts per rep. So reps territory on average was smaller. And that was by a design. And these, we moved to much more industry standard metrics in all of the changes that I'll just describe here. And we're designed to help our sales teams go deeper into those accounts, penetrate them further, grab greater share of wallet. And it's the right enterprise selling motion, and we moved to a more industry standard norm there. Similarly, in terms of our broader enterprise sales coverage, again, similar concept where while we added some more enterprise sales coverage, we actually reduced the number of accounts that sales reps carry quite significantly, which meant that people can't go to the lowest hanging fruit. They actually have to go deep into those accounts, build those relationships, and that will allow us to represent the portfolio more fully to those customers and drive higher adoption across use cases. And then the third thing we did was we created new greenfield territories as well, and these were accounts where we may not have had a lot of business historically and they represent net opportunity for us. If you think about the total addressable spend and how much those customers actually spend on various different solutions in these -- to solve these kinds of problems, it represents opportunity for us. And that opportunity would take some time to cultivate. So those were the big changes that we made. I'll point out that these were mainly in the Americas outside of -- and particularly the United States, we did not segment public sector because that has its own -- or I should say we did not resegment public sector because that has its own selling motions within federal and also state and local education. And outside of the United States, there tend to be just natural geographic ways in which markets are segmented. A salesperson in France usually won't call on accounts in Germany, for example. So within Europe, you have these natural segmentation limits that occur anyway. So most of the changes for us were here in the United States. And what happened for us was the ranges were bigger than what we had historically done. And as we implemented those changes, I think where we came up short was on the way those changes were implemented over the course of the quarter. And how sales rep -- one sales managers hands off to the other sales rep, the incentive structures you put along those, just things that you do to try and make that more effective. And I think that's where we came up short. So the impact of that was really felt in the month of July and particularly the last few weeks of the quarter because as you think about the enterprise software landscape like many other enterprise software companies are quarters tend to be heavily back-end loaded. And when we made the changes in the month of May, we started to really see the effects or the impact of that only towards the end of the quarter. And by then, it was too late to actually change anything within Q1 itself, but we obviously started to put significant actions in place to address that. I think those -- we are seeing really encouraging signs of those actions taking hold here in Q2. Many of the deals that had moved out of the quarter have already closed. We've not seen any meaningful shifts in our win rates. So we don't believe it's a competitive issue. The other deals that remain in the pipeline are stated to close in due course and this quarter, some in future quarters. I think that was the biggest impact associated with the segmentation. The other thing we saw on this was to a much lesser degree was towards the end of the quarter, we saw some deals push out in EMEA in particular. And I think that was more because of specific budget issues in those customer settings. I don't know that it's broad enough that I would call it a deeper macro issue. We don't have enough scale to be economic bellwethers ourselves, at least not just yet, someday we may. But I think we saw a little bit of that, which impacted the quarter as well. So we're on our way to addressing the issues and hope to have them address in the next couple of quarters.

Yes. So I guess going back to the change in I think as you described also in the earnings call, this was a pretty broad-based change that impacted almost every rep in the U.S. and not only the resegmentation, but the introduction of these new greenfields. So I guess when you initially set guidance, did you just think that this was going to be a pretty smooth transition? Or what was sort of the underlying assumption for this transition? And I guess beyond the booking shortfall at the end of the quarter, what were like the signals that you saw that this transition was going to be a bit bumpier than you had planned?

Yes. Clearly, we anticipated some effect associated with it. And in terms of our approach to guidance, it's been very consistent of the quarters where we try and incorporate everything that we know into the guide and then try and protect ourselves to a degree for things that we don't know that may come our way. We took a similar approach and we protect -- tried to protect in the guidance to some extent for this issue. The problem is that it turned out to be a much bigger issue than we could have anticipated. And it showed up much later in the quarter. And that's part of the reason why there was a more immediate impact from a guidance perspective, looking out towards the rest of the year. In terms of the other part of your question around signals. I think the best signals to look at are deal progression through the pipeline. Because so much of it was slated to close naturally in July, we didn't see some signs of that earlier. But as we've continued to examine exactly where those deals are and Mark is deeply involved in this. Ash is personally engaged in this, even during the break over here have been dealing with a couple of deals and making sure that they progress nicely here through Q2. Like we're all down at the individual transaction level for the transactions that are more meaningful are working on making sure that we are making adequate progress on those. And that's the best indication. As I mentioned, some of those deals have already closed and that's a really positive indication in other settings, the feedback from customers is that we weren't there driving the transaction forward with them. It is not like those deals have gone away or it's not like they've been lost to competitors. So we're still working those and are working to try and bring those to closure.

Okay. I mean, in some ways, it sounds like the customers should be even more involved with -- or sorry, the sales reps should be even more involved with customers if they -- accounts to focus on. So I guess in terms of the specific steps that you're taking to remediate this issue and there's only one of you and one of Ash, which I'm sure that's helpful getting involved. But what are you doing to -- like what were the solutions that you sort of identified to remediate this issue?

Yes, there are a few things, right? One is just naturally continuing to engage with customers, engage with the accounts. I think the second one that just happens quite naturally to be honest, is that, so it's less of an action, but more just making sure that it's actually happening is that as sales reps settle into those territories, they're naturally uncovering new opportunities. We are seeing that on the enterprise side as well as people engage with the new accounts. You're seeing greater opportunity discovery. You're seeing greater customer engagement in terms of meetings, in terms of customer participation at events where we get called in as executives to engage with those customers as well. And the third is simply just the passage of time as people become familiar with their territories. If you had somebody that inherited a set of new accounts and it's going to take them a while to build those relationships. We are starting to see encouraging signs of that happen as well. I think in addition to all of those things, there were a couple of things that we were planning anyway, launching, and it was not necessarily directly a result of the impact of the segmentation. But those are things that were in the works anyway, and we have done, and we are seeing really encouraging signs of that. The two things I'll call out specifically on that front was we launched in what we call the Elastic Express migration program. We launched that at Black Hat. We've talked about platform consolidation as a theme for a while now where customers are looking to move off legacy vendors, especially on the -- for SIEM and logging and move on to Elastic. And what we did was, if you think about it from a customer's perspective, they're in that migration, trying to minimize costs and minimize risks. And we've helped them minimize risk from a technical perspective through the product in a number of different ways. ESQL was a great example of a feature that we launched more recently. We launched automatic import for SIEM, which makes the migration a little bit easier. But commercially, what we did was took a bundle of services, which would -- in order to help the customers through the migration as well as provided them with some relief from dual vendor costs during the migration period, to construct offers that help them eliminate that friction in the near term, but they are still commercially sensible for us as a company. And we package that up and launched that as the express migration program. And that's been very well received, not only by our field, but more importantly by customers who see that as lowering the barriers to adoption. Another example of that was in addition to all the other segmentation changes we made, we at the start of the year, incubated a very small team of technical search specialists. We had some success with that in security in the past. We did that with search now as well. And these are individuals who work with customers, who are working through their Gen AI architectures, thinking about their applications occasions, and they're helping those customers along that journey in terms of how best to use Elastic and our capabilities on that front. And there's so many demands on that theme's time. It's really been encouraging to see. So there are some natural things we are doing anyway to help eliminate the friction and drive execution as well.

Okay. Okay. Got it. So obviously, with anytime you have these unforeseen shortfalls, they can be pretty painful. But I think what investors want to get comfortable around is how do we not get surprised again. So maybe as you think about the guidance update, which did have a decent cut to guide even despite a bit of a beat in the quarter. Can you talk about the assumption that you embedded? Are you assuming that this suppressed bookings environment continues throughout the full year? Is it resolve itself by Q2? And then I'd also be curious on the EMEA point that you made which does, by the way, line up with even how Microsoft talked about Azure softness. Is that something you're seeing continuing here in the Q2 as well? And have you sort of downgraded EMEA expectations for the rest of the year?

Yes, if I think about the broad assumptions, maybe starting with the macro and then talking about deal closure rates and so forth. In general, we've assumed that the macro environment stays where it is. And that includes EMEA as well. We are obviously mindful that some deals took a little bit longer to close. And as I said, I don't know that, that is necessarily a broad macro issue from our perspective just yet. But we have been cautious in terms of how we think about that part of the business as we've looked at the guide. In terms of the other assumptions that go into the guidance, we have assumed relatively prudent close rates on the pipeline and in terms of just our ongoing sales execution. The issues that we experienced and the fixes that we put in place, although we are seeing encouraging signs of progress here in Q2. We have assumed that they will take a couple of quarters to come back to our normal base execution. And in the guidance, continuing that philosophy of guiding based on what we know. We've taken everything we know about what happened in Q1, including the current state of affairs and the sales organization, the work that we are doing to address that. And that's all being factored into the guide. As I said, we've assumed relatively proven close rates. And then beyond that, again, we've continued to protect ourselves against known, unknowns to a degree.

Right, right. Okay. Okay. That makes sense. One of the other topics that I think maybe got overlooked at earnings, given all the moving pieces on the guide was the announcement of the return to open source AGPL licensing for kind of the core Elasticsearch product.

Super excited about that.

Yes. Yes. So I'd love to talk about that. I mean, what -- I guess, what sort of prompted the move at this point? Like why now? And I'd imagine you'd love to make that announcement under better circumstances with go-to-market issue. But yes, would walk us through it.

Yes. It was clearly unrelated to anything related to one quarter. For those of you that may be familiar with Elastic or have followed the story for a long time like you have Tyler. You know that we have a lengthy history. We were founded -- our roots are an outsourced and we have founded, we used to offer a lot of our software under open source license. And over time, what happened was, we saw that Amazon was offering a completing cloud service. They used to call that back in the day, they used to call that the Elasticsearch service. And a big part of our license changes at that time were motivated when we a few years ago, we said we're not going to publish our free software under an open-source license anymore. We still made it available for free. We still made it available as open code, so you could still inspect the code and play with it. But instead of being -- instead of the source code being licensed under open source, it was licensed under a proprietary license or under the SSPL license, which is, again, a relatively permissive license that not technically open source, but looks and feels very much like it. And the intent of the change at that point a few years ago was to make sure that we are continuing to protect our IP and that Amazon eventually went a different direction. We knew they would fork the code of the old open source version. We also settled the trademark dispute with them and that became the birth of open search from Amazon. So we felt really good that Amazon has, now for the last 3 years or so, gone down this path of open search. And over those 3 years, we've also built a very healthy and vibrant partnership with Amazon. And you see that in a number of different ways. You've seen us win awards, you've seen a lot of traction for us on the AWS marketplace. You've seen a lot of technical integrations being built with Amazon. So the partnership is in a really good spot right now. I think the other thing is all of the hyperscalers have done really well in terms of thinking about a first-party service and a third-party service, both running on their underlying infrastructure. And they are used to managing that as a concept. And so we search really good about where we are from that standpoint. And so in some senses, returning to open source elsewhere a little bit like a return to our actual rules. And more specifically, what we did on that front was we took the free and open version, the version that was open code and put that under not only our Elastic license and SSPL, but we made the source code now available under AGPL as well, which is an OSI endorsed open source license. The important distinction to just keep in mind here is that nothing that was free has gone into -- excuse me, nothing that was paid has gone into free as a result of this change. The only thing that changed that what was already free continues to be free except that under -- instead of being available only under the Elastic license or SSPL is now also available under AGPL. And that unlocks significant opportunity for us on the vector database side, going back to the timing of why now. That couple of reasons were, one is we felt that we had enough distance that the competitive issue had been addressed. But importantly, it's about the opportunity in the vector database side. And because AGPL is an OSI endorsed open-source license, as you think about vector databases and open source vector databases, it suddenly unlocks a whole set of opportunity for us on that front.

Got it. So you sort of feel like the maybe bottoms-up grassroots developer adoption of the vector database was maybe it wasn't right where you want it to be if that's the right way to think about it. But this sort of is an accelerant to that.

It is an accelerant. And I think the adoption that we had was very strong and you see that in a number of the statistics that we've been printing in every earnings call in terms of customer adoption of our vector databases. You can see that at our user conferences where we conduct these tour stops around the world where you see a very strong degree of engagement. But being open source just unlocks even greater opportunity for us.

Got it. Got it. Okay. Shifting to the products and the Gen AI stuff because I think a lot of that is sort of was we were distracted by the go-to-market issue last quarter. But just for folks in the room, can you sort of just articulate what Elastic offers in terms of Gen AI, ESRE is often used. I think there's kind of an umbrella term to describe a suite of products, but what is your approach between vector database is different types of search?

Yes, it's a great question. And you're right. As raised that of an umbrella term, the key word there is really about relevance, which is the R in the ESRE. And if I think about all the features, ESRE is not a product just for those that might not be familiar with our product portfolio and how we go to market, what a customer buys is the Elastic stack and depending on which tier they buy subscription tier, they will get different levels of features across all of the different solutions. And if you think about ESRE, it's not a stand-alone product that is sold separately, but it's a collection, it's the umbrella term for a collection of features that relates to Gen AI and Vector Search is one of those. We have our Sparse EncodeR model. We have an inference API service. There's a number of different features that customers use either simply for vector search or in order to build Gen AI applications that all come within that ESRE umbrella. So that's what we have there and we help customers across all of these. If I think about the rate of customer adoption that we've seen. We've seen every quarter for the past several quarters now we've seen hundreds of additional customers adopting us for vector search solely in the cloud, right? When we talk about more than 1,300 customers now, those are all cloud customers. There's a larger number of customers that are using us in self-managed mode for vector search. Similarly, if I think about the pool of customers, that's the customers that are more than $100,000 in ACV in the aggregate, which is our proxy for some of our larger customers within that universe, there's more than 200 customers using us now for vector search. We talked a little bit this quarter about a slight shift towards search bookings that we saw. So we're really starting to see these things come together quite nicely.

Got it. And as you think about the competitive environment within vector search and Gen AI, right, you have a number of private companies in the space, be it Pinecone or Weaviate. You also have vector database offerings from companies like MongoDB or even Postgres. How do you sort of see yourself like what type of use case. Is it anything around search? Are you trying to kind of go after the broader vector database market too for Gen AI applications?

Yes, I'm glad you mentioned Postgres because we actually think they have implemented the right way in by doing this directly in the query engine. But stepping back a little bit more broadly. I mean, fundamentally, we view ourselves as a search engine. And from our standpoint, there's a number of other players out there. You can always have stand-alone tools where you've got some amount of data that is sitting in the tool itself. And you can have some search functionality that is resting on top of that, which will be inherently limited to the data that is in the tool because it doesn't have visibility to data more broadly within the enterprise. And I think those will continue to exist. We fundamentally don't see them as being competitive to the way we approach the market. And from our perspective, searches is much more ubiquitous. With respect to some of the stand-alone players like the Pinecone or Weaviate and so forth that you mentioned. Our view for a long time has been that ultimately, vector search becomes a feature. It's not a product category by itself. And that's a big part, the reason why when we implemented vector search, we implemented it within Lucene, and we implemented it natively into the Elastic stack. And back then, we used to have some FUD thrown at us around performance, but that has long been addressed. And if I think about what now differentiates us and how we see ourselves winning to start with. If you think of vector search as a feature, you have to have high performance on that feature. And we compare very well with any of the others out there. In terms of popularity, we tend to be much more popular simply given the size of our installed base because we implemented this within the stack itself. And now with our vector database being open source effectively under AGPL for people that want to use that. It becomes an even greater accelerant like we were talking about. The other important piece is that in order to succeed from a technical standpoint, we've maintained for some time that customers will need a lot more than just vector search itself. They will need all the other features that enterprises have become used to when you think about implementing search. That includes a ton of features around things like security aspects. So just features that enterprises would come to naturally expect over time. And we have a long history of delivering on those features. Even from the standpoint of economics, we talked in the past about examples like hybrid search and reciprocal rank fusion, which was the technology to enable hybrid search, which allows customers to get the best results across traditional lexical search and vector search without rerunning the search and to rerank the results. And you can't do that unless you have access to all of the feature sets that we have. At least can't do it as effectively as we can. So we feel very good from the standpoint of our competitive positioning. I think over time, you will continue to see more innovation, you always do in the technology landscape. And we feel really good about how well we are positioned with our offerings for customers. And you see that in the numbers, a number of customers that are adopting us.

Yes. And your comment earlier where you sort of saw more of a mix shift towards search, at least in terms of the bookings. And I think you talked about on the call that search was actually reaccelerating in terms of revenue growth, perhaps outgrowing the total company business. Is that all being driven by AI? And I guess, how close are we to the ESRE family of products really starting to move the needle in terms of driving revenue?

Yes. I think it's largely being driven by Gen AI. But in terms of when we start to see that from a revenue perspective, one of the things that we talked about when we entered the year was that we -- while we had seen some revenue contribution from Gen AI in fiscal '24, we were not expecting to see meaningful inflection in fiscal '25 because we do think this is a longer-term play, like most major technology transitions, we think this takes longer to happen, but ultimately ends up being bigger than what many people anticipate. And so we think of this from a longer-term perspective. And that view hasn't fundamentally changed. On the one hand, we are really encouraged with what we saw in Q1 with the additional commitments that we secured and the customer interest translating into this higher search mix. But over time, I think that's when we really see a more meaningful revenue impact. Mix shifts are also really based on relative growth rates and can bounce around. So we're not trying to engineer a particular mix shift in the business. But so far, really encouraged by what we've seen on Gen AI, but we just think it's longer term from the standpoint of how this plays out on the revenue side.

Okay. I know we've spent a lot of time talking about search in ESRE and everything, but I did want to talk a little bit about the other businesses, too, which are just as important. On the logging and SIEM security analytics side, what have you seen thus far, just with Cisco now I think owning Splunk for 2 quarters now, if I'm not mistaken. How are you sort of positioning your team to go after that opportunity? And have you seen any potential benefits or any positive signals that you're seeing in terms of being able to go out and capture that?

Yes. I think when the acquisition first happened, our view was that the reason a customer is concerned that wants to move away from legacy vendors is usually because of costs or because of lack of innovation. And those were the main reasons why people were thinking about adopting Elastic and coming to Elastic compared to some of these legacy folks out there. And that fundamentally hasn't changed even post acquisition in relation to Splunk. So we've not seen any big shifts or dynamics play out in front of customers as we are engaging with them on particular opportunities. And we've continued to feel really well positioned on that front. I think customers continue to give us the same feedback that they've always given us about the reasons that they're adopting Elastic relative to some of the legacy players. And our goal over the coming time frame is to continue to try and accelerate that movement. And that's part of the reason why we adopted -- why we launched the Elastic express migration program features like automatic import, which is a relatively recent feature, will continue to help. So we feel really good about how well we are positioned and the future potential on that.

Okay. Okay. Great. And then on security, I'd be remiss if I didn't ask you just post the CrowdStrike outage and all the fallout on that. I mean, I guess a couple of perspectives. Number one, did you see any material share gain opportunity in terms of Elastic endpoint products and then did that sort of trigger more logging use cases or security analytics use cases that you saw during the quarter?

Yes. There was no direct impact. And the CrowdStrike issue, I have a lot of respect for them as a company. They've obviously built a great company over the years. But we're not directly competing with them on endpoints. Our focus on security is primarily on SIEM. And the endpoint issues that they saw really didn't have any direct bearing on what that means for us from a SIEM standpoint. I think customers are very aware of our offerings and what we bring and how we show up. And I think that plays nicely to our advantage overall.

Okay. Okay. Got it. One of the product announcements that did not get as much airtime this call, but it's sort of coming down the pike. And I know there's a lot of excitement around it is with Serverless. And so I was wondering if you could sort of just talk about where we're at with that, what the benefits are for customers and then how investors should think about the potential pricing changes and how to think about the rollout of that across your customers?

Yes. So it's a great area and one that has been a high priority focus area for our engineering team for some time now. Fundamentally, one way to think about Serverless is what it lets you do is it lets you decouple the compute from the storage. And what that does is a few different things. One is it unlocks a whole new set of use cases. If you've got use cases that have particularly bursty needs in one dimension of those. It allows you to scale those somewhat independently. But importantly with Serverless, it actually changes a lot of the actual user experience. And it's a lot simpler and easier in terms of thinking about what that customer experience looks like because, the experiences are more bespoke to the particular solution areas. If you're a security operator, if you're DevOps person and you're looking at log analytics, if you're just thinking about search, historically, those were all -- the experiences that we had in our traditional hosted product were somewhat catered to those. But this is a true SaaS-like experience in terms of thinking about not having to deal with a lot of the orchestration work that you otherwise would have had to handle, not having to deal with a lot of the management work, things like managing, shorting and the size of your indices and so forth. So it actually simplifies the management quite significantly, and it makes it a much, much more -- much better user experience. And the third thing is from a pricing standpoint could actually allows us to have pricing that is more specific to the solutions or to the use cases. And that actually helps us in the longer term from the standpoint of thinking about our gross margin where we can price based on a particular use case and the pricing is connected to -- indirectly connected to but not directly informed by immediate changes in the pricing of infrastructure. So it gives us more flexibility from that standpoint as well, and it gives the customer more flexibility as well from the standpoint of the way they think about it. So it's been incredibly positive on all of those fronts. Serverless is currently in beta, and it will go GA by the end of this calendar year. For us, that means we need enough presence in the different hyperscalers in the geographies. But so far, the reaction to the service has been incredibly strong. What's important to know is that Serverless is not going to replace the existing Elasticsearch offering. So we're not going to do any kind of forced migrations or anything of that sort. It will be offered alongside the existing hosted offering. And so we think a lot of new customers and new workloads will go to the Serverless offering. It might be some people that are more comfortable with the existing offering. Particularly if they've already got all of their processes set up, their dashboard set up, if they are comfortable with the pricing model, they may stay there. So I think over time, we will naturally see a shift to Serverless but we are not trying to engineer a particular mix shift there.

Right. And just thinking about the financial impacts of that. Obviously, tough to predict because it's in Beta. But just with the separation of compute and storage, is there any risk of revenue loss if there's more efficient storage or conversely, maybe the Serverless pricing is going to be at a premium. And so if there is conversions, that turns to price tailwind?

Yes, I don't think there's any sort of immediate risk of big revenue hit associated with it because as I said, it will also take time for it to ramp quite naturally anyway. Net-net, I think this will actually be positive for us, both in terms of aggregate revenue dollars as well as in terms of gross margin percentage.

Okay. Okay. Very clear. I guess on margins, we didn't talk too much about that. But despite the top line shortfall, you actually did take up margins. If you could sort of speak to the actions that you did? And then philosophically, how do you sort of think about long-term growth and profitability at Elastic? Or do you aspire to be a Rule of 40 company. How do you sort of think about that?

Yes, we've not formally adopted a particular framework but conceptually, that's exactly how we think about it, that you have to balance revenue growth with profitability. And the right to keep investing in the business is earned based on the results that you deliver on the top line. And these models tend to have enormous leverage on the way up and down. And so the way we've tended to think about it is, when we see continued growth in the business, we'll invest towards that growth while we're continuing to expand operating margins at the same time because there is leverage that is inherent in the business model. And if growth slows for whatever reason, then we step back and assess why did that happen? And if it's something that is short term, we take actions in the short term. If we think there is some structural shift in the market, then we will take appropriate actions based on that. But fundamentally, it is about making sure that we are driving for both revenue and profitability and achieving that balance. So that's what you saw us do this time as well. If I think about fiscal '25 and the revenue guidance, we brought it down by 2 percentage points in terms of year-over-year growth. As you think about that change in the revenue guidance at the midpoint, that was about $34 million of total revenue at the midpoint that we brought down. And we brought down our expenses by a little bit more than that and still protected and slightly increase the op income in the guidance. And the way we did that was by cutting back on certain areas of investments that were planned that don't necessarily have a direct impact on long-term growth and that includes being a little bit more targeted in the sales function in terms of investing in selling capacity only in those places where we see the stability of execution. From a long-term perspective, staying the course on Gen AI investments on the R&D side, we are slowing down some of our investments in non-selling roles in the sales organization, slowing down some of our investments from a marketing perspective, particularly on the brand side and close to my heart, continuing to drive efficiencies on the G&A side of the organization. So there are things that we've done to take steps that I would call proportionate and measured based on what we saw as a short-term execution issue.

Great. Well, I think we are out of time, Janesh, thank you very much. There was a lot of detail. I appreciate the explanation. And have a good rest of the day, everyone. Thank you.

Thank you. Thanks again for having us.