All right. Good afternoon, everyone. I'm Sanjit Singh. I run the infrastructure software coverage on the Morgan Stanley software team. Super pleased to have the management team from Elastic. We have CEO Ash Kulkarni and CFO Janesh Moorjani. Thank you both for coming to the conference.
Thank you for hosting.
Thanks for hosting us.
All right. Let me get through some disclosures, then we can start the discussion. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If any questions, please reach out to your Morgan Stanley sales representative. To maybe kick off the conversation, Ash, just to start with you. If you look at the three core sets of opportunities that you're going after in Security, Observability, and Search. As you look at where your customers have been and where they're going, I guess the high- level question is: How early or mature are you in each of these core opportunities? In terms of the kind of where do you sort of stack rank on the priority list, going forward, how would you sort of frame that out for us?
Thanks. As I look at Elastic as a whole, fundamentally, I think of us as a search analytics platform. At the core, we're a data platform. What we're really good at is bringing in any and all kinds of data into the platform really quickly. We do incredibly well when it comes to unstructured data, which is quite messy. Logs, any kind of unstructured data, we're able to index it, make it all searchable. We have pretty rich, developed set of machine learning capabilities that allow you to run all kinds of algorithms on that data and then visualize that information. The use cases that we always tend to play in are ones that play to those sweet spots, right?
Involves a lot of unstructured information, involves being able to deliver results in real time, and that's the reason why Security, that's the reason why Observability. As we look at the spaces that we play in, obviously when it comes to Search, enterprise search, that was the genesis of the company, so that is an area where we are very well known. You look at the Gartner Magic Quadrant for Insight Engines, that's the space. You know, they consider us to be a leader and sort of the farthest most to the right on the x-axis. When you look at Observability, where we tend to land, our core sweet spot is in log analytics. We always start with log analytics.
That's where we not only have a very mature solution, but in terms of our competitive win rates, they're very strong. We're able to, you know, do well both in terms of new opportunities but also expansion opportunities. Once we land with log analytics, we expand from there. Whether it's application performance monitoring or metrics or real user monitoring, we'll expand from that core position of log analytics. In security, it always starts with SIEM or Security Analytics, so that's our core landing spot. Once we are in there, then we expand from there, not just to detection, but protection, so endpoint security and more recently, the work that we've done around cloud security. That's, that's how I think about it.
In terms of maturity, you know, where we have the greatest strengths are in these areas of log analytics and SIEM, and then obviously enterprise search.
Makes total sense. I wanted to circle back to Q3 results, Janesh. You know, you had a solid quarter, you grew revenue growth, 27%, Cloud business grew 40% constant currency. You know, in terms of Q4, which is your next quarter, you're calling for high 18%, I think 18% constant currency growth. For the forward year, you give sort of an initial view of sort of mid-teens growth. Can you walk us through the underlying assumptions on, you know, in this more difficult environment that we're in, why we can more or less sustain growth, fiscal year 2024 over fiscal year 2023?
Yeah, happy to talk about that. You know, for us, given that we just finished Q3 and are entering Q4, obviously like you said, it's a difficult environment out there, relatively uncertain. At this point, we at least wanted to provide people with a general framework of how we are thinking about the future, even though we won't formally guide until our next earnings call. What we said was we expect fiscal 2024 growth to be in the mid to high teens in terms of revenue. There's a few different things embedded within that, and maybe I'll start with what's not in that. We're actually not baking any kind of assumptions in there about some improvement in the macro environment.
We generally talk about the world the way it exists today and the way we see it today. We're not anticipating any significant changes or improvement. It's a wide enough range at this point that we still have some ability to withstand variability, frankly in either direction, even if there's some level of downward pressure in that number. The other piece is that fiscal 2024 really comes down to first off, it's our own internal execution. Q4, it's our biggest quarter. It's seasonally the largest quarter for us. April will be the biggest month in the quarter for us. We had a really good Q3 from the standpoint of how well the sales team executed. We were very pleased with that. Came in well ahead of what we were expecting.
Q4 is still a big mountain to climb, and so we like to queue here in Q4, and that will allow us to form a more thoughtful view of fiscal 2024 as well. The other piece we had as well. We do look at those, and we do think that that's a good reflection of how this will all eventually translate into revenue over time.
For just a follow-up on like the customer commitments as just sort of an example. If a customer has signed up for a minimum commitment or a purchase commitment.
Yeah.
They're not ramping as fast as they had expected. How do you sort of handle that customer? Do you sort of roll over that unused capacity into another contract? Or is it more of a take or pay type construct? How do we think about customers who may be falling behind-
Yeah
...against executing their commitments?
Yeah. It's a great question, Sanjit, because, you know, consumption models are still relatively new out there, and they're evolving. You know, for us, when a customer signs an annual contract or a multi-year contract, even those multi-year contracts with annual break points, and there may be ramp embedded in that if a customer's assuming that their consumption will grow over time, it could be linear. It really depends on what the customer wants. By and large, what we've seen is that customers don't over-commit.
You know, they are more thoughtful about making sure that they don't end up with a large overcommitment and the additional points of discount that they might get for a commitment may not be worth that risk for them because our contracts are generally structured as use it or lose it contracts. So what we've generally experienced is that customers will actually tend to consume faster than the rate of commitment. But in the handful of cases where we do see customers consuming slower, then, you know, in those instances, we'll either work with them to arrive at some sort of alternative approach, or in a handful of cases, they do actually have minor amounts that are actually forfeited under the contracts.
Mm-hmm.
As I said, for the most part, I think what we see is customers actually consume faster than what they were contracted to consume for. I think the other piece, by the way, from a consumption perspective that as a dynamic that plays into the numbers is around the net expansion rate. Our on the cloud business, all of our contracts, the commitments that customers make don't show up in the net expansion rate because the net expansion rate is measured based on the actual consumption. When you've got these large contracts which are still waiting to be consumed or the consumption is ramping slowly, that adversely impacts the net expansion rate because the lower level of consumption is in there, but the higher level of commitment is not.
I think that just plays itself out over time.
I guess the point there is that ultimately those customers, the vast majority of time, get to that consumption level.
One way or another, it becomes revenue.
Yep. Makes total sense. What are the themes in the, in the broader cloud ecosystem that's been around this topic of cloud cost optimization or rationalization? We're seeing it at the major hyperscale levels. We're hearing and seeing in parts of the independent software ecosystem. When it comes to Elastic Cloud, what are the various levers that customers have to optimize their spend as they look to preserve budget in a tough environment? To what degree are you guys helping them to achieve that?
Yeah, maybe I'll touch upon that. You know, we've seen different patterns, right? For example, customers might sample some of their data. We've seen some customers retain data for shorter periods of time. Instead of a year, they might bring it down to nine months. In some cases, we've seen them move more data to low-cost object storage, like in the Frozen Tier-
Mm-hmm
...that reduces their overall cost a bit. There are a few elements to this that are important to understand, and probably the first one is that fundamentally, data is still continuing to grow, right? They are doing all these optimizations, but there's a limit to which they can do these optimizations. The way we are approaching this is rather than, you know, trying to be against this, we are leaning in and having a conversation with them on, we'll help you. If you have an existing workload on Elastic and you're looking for ways, like these are all trade-offs that you're gonna have to make. Obviously, you don't get the same outcomes if you sample the data. You don't get the same outcomes if you're storing it on lower cost object storage because the performance vector there is different.
That's fine, we are helping them with it. In that same conversation, we are also trying to figure out what more can we do for you? Clearly we are demonstrating that our pricing model is a lot more advantageous to them.
Mm-hmm
...than a lot of others. In this point in time, like the worst thing that you can do is be, you know, seen as inflexible or high cost. We are able to come in and help them actually reduce that cost and have the conversation of, "Oh, you're doing APM with some other vendor," or, "You might be using some other vendor for SIEM. How can we help you reduce that total cost-
Mm-hmm
...by moving some of those workloads to Elastic, which you are already seeing as having the ability to reduce your overall costs?" That is a big part of what is helping us bring new workloads onto our platform. A lot of the Q3 activity that you saw in terms of the strong contracts from customers reflected that kind of activity. Fundamentally, we see this as an opportunity in this current time when everybody is concerned about their total budgets, to see how we can take market share. I mean, obviously we can't control the fact that there is optimization happening across the board, so consumption is slowing down.
What we can control is the activity that we drive from a sales perspective to try and take share in this time, and that only sets us up to be in a stronger place as we come through this, however long it takes.
Yeah. On the pricing front, do you wanna just, Ash, quickly just describe Elastic's approach to pricing and how it may differentiate versus-
Yeah
...the other players in the space?
Our pricing model, we basically do not sell our products separately. There's, you just effectively are buying the platform. You're buying Elastic, and there are a few tiers, and in each of those tiers we have incremental functionality, but that functionality exists for Observability, Security, Search, all the use cases that you can drive with Elastic. Our pricing model is purely consumption-based. Now obviously if it's self-managed, where somebody is just purchasing the licenses from us. They are purchasing it on a virtual CPU basis, and then they are running it. You know, we don't have the ability to meter it on a second-by-second basis. If it's Elastic Cloud, you know, we are measuring and monitoring and metering everything, and we are charging the customer based on their actual consumption. That's the model.
What that lets people do is very flexibly try out using Elastic for additional things. A lot of times, somebody might start using Elastic for log analytics, and when we have these kinds of conversations, it's very easy for them to maybe instrument a few of their applications and try out our APM functionality or try using Elastic for cloud security. Once they see that not only are we functionally very capable and have a rich set of capabilities, but also the pricing model scales a lot more gracefully, like that becomes easy for them to then start bringing more workloads onto us.
Mm-hmm.
That's a big differentiator. You know, the value for price equation tends to be very favorable for us.
Going back to the topic of sort of cloud cost optimization, from your perspective and what you've seen at your customer base, how long does a cost optimization project initiative take, and what % of the base are sort of actively involved in these types of initiatives?
We've been seeing the optimization efforts from customers across the board, across all sizes of customers and segments of customers. You know, the moment somebody decides that they wanna do something to optimize their usage, whether they are sampling or doing something different, it shows up immediately in our metering, right? The impact is seen immediately. And at this point, we're seeing it quite broadly, so it's not. You know, even now some customers are doing it more than others. Like it's what you saw in Q3 was sort of everything balancing out. I think cost considerations are on everybody's minds right now.
Yeah. I guess the point is here that it's, you know, relatively quick to implement.
Yeah.
It sort of goes to the point is like when this is done, we're in a better environment. The spigots could potentially turn back on, more meaningfully once we get into a better budget environment.
For the last many years, budgets weren't, you know, the most important thing for most CIOs.
Mm-hmm.
Right now, their budgets are like the first and second and third thing that everybody's thinking about. There is a tremendous willingness to look at how do they do more with less, and that's the opportunity that we see. That's what we're leaning into.
I wanna combine like two of the questions that I'd written down into one. I guess big picture, sort of explain to us why does Elastic win more like more broadly. In this current environment, I think a lot of players in this space is talking about the opportunity around consolidation and vendor consolidation and tool consolidation. Can you give us the cases why Elastic is well positioned to be a sort of net consolidator spend in the categories you played in? You touched the pricing point, but if there's anything else besides the pricing point to round this out.
Yeah. I'll come back to the pricing point maybe at the end. Think about how do they want to do Observability. We have lots of customers who basically want to be able to look at the performance and availability of their application based on their end customer journey. They wanna be able to view it in a particular way. We have customers who want to do things like analyze the performance for our most important clients. That means now correlating your observability data with your business data because your most important clients are typically, you know, it's a subset of your customer base. Depending upon your customer IDs, you're pulling in some cohort. Like, if you wanna do that with any other observability product, you've got to take both those datasets out, put them in a data warehouse, and do that analysis offline somewhere.
In the case of Elastic, because we're a data platform, there is no limit to what kind of data you can bring in. We have these customers just bringing in their business data into Elastic and then doing the correlation right there. We have customers who, for insider threat analysis, are joining their physical security data, the card scans that people do with their application logs and their access and identity management logs. You know, traditional SIEM platforms just aren't able to do that, but we can because fundamentally we're a data platform. There's no limit to the kinds of data and the kinds of analytics that you can do with Elastic, and that's the second most important reason why we tend to constantly win. We talked about price.
The pricing model tends to be very permissive, and it really aids in that kind of viral adoption. Probably the last reason is we've been investing a lot in machine learning. Last summer, I talked about vector search and all the work that we've done in vector search. That makes not just search a lot more powerful, but it has a lot of relevance to observability and security use cases. That's also driving all kinds of interesting, you know, use cases and, you know, enabling us to compete even better. In AIOps, as an example, which is all about machine learning for observability, we are now considered to be one of the really strong players in that space. That's all coming from these investments in machine learning.
I think those are the core reasons why we tend to win. You know, the way I see them, it's these are strengths. These are pretty significant moats that we can continue to build upon.
One of the more popular questions I get from investors is that they say something to the end of, "I like the market opportunity. I like the product category that Elastic's in, but the market's so crowded in terms of the number of players at least saying that they can sort of compete in the, in the space between commercial solutions, open source solutions, and even the hyperscalers have their own set of functionality." As CEO, how do you think about competing in a market that does have you know, multiple alternatives in a way that'll translate to durable growth and profitability? sort of talk to like that-
Yeah.
...that sort of challenge and, you know. Is it a view that the market's so big it can support multiple players?
I think the market's big enough that it's definitely gonna be able to support, you know, several players, right? That doesn't mean that it's gonna be able to support dozens and dozens of players necessarily over a long period of time. The way we look at the markets that we play in is do we see a viable path to be one of the top two to three players in each of the core categories over a period of time and sustain that? Hence the focus on log analytics, hence the focus on SIEM and XDR, hence the focus on search. Those are areas where, you know, you just look at security. In Security Analytics, you know, like The Forrester Wave put out an article recently, we are one of the top three players.
Mm.
When you look at AIOps, we are doing incredibly well in that area. When you look at search, we're kind of like the de facto name out there.
Mm.
In the, in the places that we land with, where, you know, you need to have a very competitive offering to ensure that you can win the RFP, you can, you know, show, demonstrate really well in these enterprise accounts and win those accounts, you know, we've got tremendous strengths. The competitive mode is one that we believe will hold. Once we land, then we expand from there. The second part of this is, you know, how do we do this profitably? I think that's a really important question. To me, the most important thing there is the platform approach that we've taken. The bulk of our R&D efforts go into the platform itself.
When we talk about features like Searchable Snapshots, we talked about that as being one of the major drivers for customers upgrading to our enterprise tier. That Searchable Snapshots functionality helps all our solutions, whether it's Security, Observability, Search. The same thing for things like vector search. Vector search allows you to get better relevance results. That's true for search. It's just as true for, you know, behavioral detections and security. It's just as true for observability. A lot of our investments actually go in at the platform tier. That really helps us make sure that we are being very efficient in how we service this market. That also helps us from a go-to-market perspective because it's not like our sales team is now having to learn four different products or five different products. It's one product with multiple use cases.
The effort for a pre-sales engineer to ramp up, the effort for a account executive to ramp up is that much better because of that. You know, we're big believers in the platform approach for that reason. We believe that over the long run, it ends up being a heck of a lot more efficient. That's really why we stay so disciplined to that approach, even in the acquisitions that we do, that tend to be really, you know, technology tuck-in acquisitions.
Makes sense. I definitely wanna see if the audience has any questions. Get ready to ask your questions. I do have to get to the obligatory AI question first, though, right? It's like can't do a conference in 2023 without talking about AI. It's a two-parter. The first part is, as it relates to your enterprise search business, do you believe things like OpenAI and ChatGPT-esque, like capability-
Large language models like the one from OpenAI or even what Google has, they all depend on something called embeddings that allow you to create like pretty long vectors with lots of dimensions in them. We delivered that support on our platform as well, so you can now bring in these kinds of models. Most customers, let me put it that way, wouldn't want to run like a ChatGPT, like the OpenAI model on their own infrastructure. It'll be just prohibitively expensive in terms of the compute cost that you would need. You know, my expectation is that most customers will have their own models that they will then augment by making API calls to something like a OpenAI or Google in the future.
I think that's gonna be a really interesting way to, you know, make your search queries that much more interactive, that much more interesting. You talked about enterprise search, but I believe that there are applications of this even in observability and security. We've been investing in this area for a while. We are not gonna be building the models ourselves. Like that's, you know, OpenAI is doing that. Google is doing that. Facebook is doing that. We are gonna make it possible for you to augment your own ML models with whatever you want to get access to from these large, you know, transformer models.
Great. I'm gonna skip my second part of the question, and let's go to the audience. There's a gentleman here right in the front.
Ash, Janesh, I appreciate the time here. I appreciate the conversation. I actually wanted to maybe bring a few topics that we've discussed here together. I know we went through the customer commitments, which are very great to see, optimization around costs. I know that's, I guess just the reality of the world that we're in, and then our pricing model. I know we touched on those three. I guess, the question I'm thinking about, I know in the, in the way we price, if customers commit ahead event, they can get better pricing per hour. In like a world of optimizations, does, you know, when customers commit more, does that help them basically get more pricing per hour?
Is that possible to think about how customers are commit or why, one reason of why we've been successful in commits as well?
Yeah. Maybe I'll start. You can add to that as well as you like, Ash. You know, fundamentally, if a customer makes a larger commitment, they do get better pricing than a customer that makes a smaller commitment, right? I think that's fairly straightforward. I think the important piece for us is, as customers think about making these larger commitments to Elastic.
What they're really doing is moving additional workloads onto Elastic, and in many instances, those may have been workloads running on competitors. The way we think about it's actually increasing our footprint overall in the account. From a pricing perspective, what we have not seen is any sort of undue pressure based on deal bands and so forth. It's not like we're giving away the farm to attract greater commitments or anything of that sort. In fact, if anything, the sales team was really well-disciplined in how we executed in Q3 in terms of protecting ASPs and deal sizes and discount rates and so forth. ASPs actually got bigger compared to the year-ago period, as well as compared to the prior quarter.
Yeah. Just to put that in a little more context as well, right? If you have a customer that's, you know, running at a run rate where they consume up to, let's say, half a million over a period of a year, you know, if that customer is looking to get a significantly larger discount for committing to $600,000, like, we'd never do it. Like, that just makes no sense whatsoever. Like, economically, that would not really, like, make any sense for us, so we would never do it. Now, if that customer commits to $5 million a year, they're not doing it for the discount, they're doing it because they can actually burn $5 million a year, because as Janesh said, these contracts are written to be, you know, use it or lose it.
That's when, you know, they are intending to bring some additional workloads onto our platform. In today's environment, if they're bringing some additional workloads onto our platform, it's because they're taking that from somewhere, right? They're taking that from somewhere who might not be giving them the same kind of economics, overall benefits, whatever, as Elastic is able to give them, and that's really how it plays out.
Unfortunately, we have to stop it there. Thank you so much, Ash and Janesh.
Thank you. Yeah.
A great conversation. Thanks.
Thanks again for hosting us.