Elastic NV (ESTC) 2019 Q3 法說會逐字稿

Good day, and welcome to the Elastic Fiscal Third Quarter 2019 Financial Results Conference Call. (Operator Instructions) Please note, this event is being recorded.

I would now like to turn the conference over to Anthony Luscri, Vice President of Investor Relations. Please go ahead.

Thank you. Good afternoon, and thank you for joining us on today's conference call to discuss Elastic's third quarter fiscal 2019 financial results. On the call, we have Shay Banon, Founder and Chief Executive Officer; and Janesh Moorjani, Chief Financial Officer. Following their prepared remarks, we will take questions.

Our press release was issued after the close of market and is posted on our website, where this call is being simultaneously webcast. Slides which accompany this webcast can be viewed in conjunction with live remarks and can also be downloaded at the conclusion of the webcast on the Elastic Investor Relations website, ir.elastic.co.

On this call today, our discussion may include predictions, estimates or other information that might be considered forward-looking statements within the safe harbor provisions of the U.S. federal securities laws. While these forward-looking statements represent our current judgment of what the future holds, they are subject to risks and uncertainties that could cause actual results to differ materially. These risks and uncertainties include those set forth in the press release that we issued earlier today as well as those more fully described in our filings with the Securities and Exchange Commission, including our prospectus and 10-Q as filed with the SEC and forms 8-K and other filings we make with the SEC from time to time.

You are cautioned not to place undue reliance on these forward-looking statements, which reflect our opinions only as of the date of this presentation. Please keep in mind that we are not obligating ourselves to revise or publicly release the results of any revision to these forward-looking statements in light of new information or future events unless required by law.

In addition, during today's call, we will discuss certain non-GAAP financial measures. These non-GAAP financial measures, which are used as measures of Elastic's performance, should be considered in addition to, not as a substitute for or in isolation from GAAP measures. Our non-GAAP measures exclude the effect of our GAAP results of stock-based compensation, amortization of acquired intangible assets, acquisition-related expenses and non-GAAP tax rate adjustments. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP measures, in the press release and on our Investor Relations website in the slides accompanying this webcast. The webcast replay of this call and slides will be available for 2 months on our company website under the Investor Relations link.

With that, I'll turn it over to Shay.

Welcome, everybody. It's great to be here today and share with you the results of our third quarter. This quarter, we saw continued adoption of our products and features, expansions to new use cases, growth across all geographies and increased mind share at all levels of business. For that, I'd like to thank our wonderful community of users, customers, developers, partners, investors and our employees and their families.

In Q3, revenue grew 70% year-over-year to $70.8 million. We had more than 7,200 subscription customers at the end of the quarter, including over 380 with annual contract value of more than $100,000. And our net expansion rate was over 130%, which we've maintained for 9 quarters in a row now. But before Janesh dives into the details of our financial performance, I'd like to talk a bit about some highlights from the last quarter, and I thought I'd start with the topic of airport security lines.

We are a distributed company, so airplane travel is not uncommon. In the quarter, I traveled to Israel and Japan to spend time with our communities and new teams there. I went to New York for our Elastic{ON} Tour event. I also spent nearly a full day afterwards speaking directly with our customers in the city. I also flew to Amsterdam to celebrate the move to our new office with the original founders, the team there and our local users and customers.

So throughout all of my travels, there was this constant, airport security lines, which made me think of Canvas. I talked about Canvas on our last call. It's a free feature in Kibana that we previewed in our 6.5 release. And I'm particularly excited about it because it shows how we built search technology that really unleashes the creativity of our users.

One place we've seen this is at the London Stansted Airport. Their security operations team wanted to use Elastic to visualize data like train arrivals to the airport, plane departures from the airport, passengers flow through security lines and even counts of those plastic trays you put your laptop in. So they started with the usual Kibana dashboards, but they wanted more freedom to illustrate the real-time flow of passengers through the airport in a way that reflected the physical space, so they turned to Canvas.

They created several work pads. One of them actually mimics the physical layout of their security lines. You know how those 2 lines sort of merge together as they approach the body scanner? That's what they created in Canvas. And the lines change color, so green for a good flow, orange to indicate a slowdown and red is when I show up. Now they have the -- this fantastic way to communicate to security staffers and even their Head of Business Intelligence, what the real-time flow of passengers in the airport looks like. Are inbound trains delayed? Can they anticipate an influx of passengers? Should they prepare to open up more security lines? Are they running out of those plastic trays? Yes, that is always the case.

Now I won't go through their whole evolution. You can learn more about that on our website, but the end result was rather remarkable. I wanted to start with this story because it is use cases like these that speak to our continued momentum across different vectors. One is broadening the consumption of our products to more types of users and use cases; the second is creating critically valuable proprietary features like Canvas, which in this case is free under the Elastic license that only we can provide; and third is the ability to store more data and stay fast while still being efficient.

I'd like to talk a bit about that third one because there's an inherent tension between speed, storage and efficiency. In logging and metrics use cases, there's often large volumes of data, and it's common to store that data in time-based indices, turn index per day for example. And as more and more data gets ingested, it can add up and become more expensive to store and difficult to manage. Broadly speaking, many technologies make it a zero-sum game. You either sacrifice speed in order to store more data or you sacrifice the amount of data you can store to maintain speed and save on cost.

But we think about that tension differently, we strive to balance it. And we're uniquely positioned to do so with the features in our products. Hot-warm architectures and data rollups are 2 examples of that, which I spoke about in our last call. This quarter, we released new features that continue to build on this theme.

The first is what we call frozen indices. This is a feature relevant to data-retention use cases, especially where auditing, compliance and long-term data analysis are important. We built this feature in order to give users the choice to more efficiently store more data, but at the expense of query speed. This is hard for us. We jokingly tagline this effort would make it slow. So now a single node can easily manage thousands of frozen indices with very low overhead. And a query that would take 100 millisecond on a hot index only takes 1 second on a frozen index. That's pretty mind blowing. Even when we're slow, we're fast.

Hot, warm, cold, frozen, if you think about it, data sort of has seasons to it. It starts out fresh and it's frequently accessed on hot nodes. As time passes, it ages and moves to warm, then cold, frozen and if not needed, deleted. And there have always been ways to manage these life cycle in Elasticsearch through scripted automation or additional plug-ins. It's just an API call away. But we knew we could make this a more efficient and integrated experience in our product. So we previewed index lifecycle management or ILM in our 6.6 release. Now a user simply defines their data retention policy, and ILM takes care of the rest. I'm happy to say that these features, like Canvas, are also free and proprietary under the Elastic license. I'm excited about them because of their impact on so many of our users, especially with logging and metrics use cases. This is just another step forward to allowing our users to efficiently search and store any type of data in Elastic, regardless of age.

Take, for example, the French auto manufacturer group, PSA. The company behind the brands Peugeot, Citroën, Opel and Vauxhall. They renewed their business with us in the quarter. They use Elastic to get real-time insights from all their logs and metrics at all levels, from engineers to management. Their data streams in from many sources: applications, infrastructure, network, security tooling and even the factory floor. And each data type has its own lifecycle that needs to be managed.

Another logging example is La Poste, the French postal service, who also renewed business with us this quarter. With Elastic, they've scaled out their centralized logging service and are capturing data from their applications, infrastructure and network. I also had the pleasure of meeting with them at our Elastic{ON} Tour Paris event, and they are excited about deploying our APM solution, expanding upon their existing logging and metrics use cases.

So in the context of these new features, I am confident our users will be able to more easily store and access even more data in our products at a lower cost. This also makes it even easier for customers to expand with us and adopt us for new use cases, which is often how people grow with us.

An example of this is Groupon, an e-commerce marketplace who's used Elastic to power the search experience for some time. Recently, they shared with us that they, and I quote, "made a strategic decision to start the replacement of Splunk with Elastic," dramatically expanding their use of our technology to power their global enterprise logging platform.

Another example is Choice Hotels, a large hotel franchiser headquartered in the U.S. who closed new business with us this quarter. They're using Elastic for application and infrastructure logging and switched to our technology for a few reasons: first, the powerful features we provide like security and machine learning; second, the flexibility to expand to new use cases like metrics and APM; the third is access to our world-class support experts.

So you can see a pattern here. We continue to see momentum behind adoption and expansion of our technology for more use cases and audiences. But this theme of speed, storage and efficiency reaches beyond just logging and metrics use cases. Take the security space, for example. Every log, every security event can be an important fiber in the larger fabric of an organization's knowledge base. Excluding data because of cost can result in holes or vulnerabilities. Elastic doesn't make users choose what to leave behind. We continue to see us get adopted in favor of other solutions for these reasons.

One example is Ricoh Group, a Japanese multinational imaging and electronics company who recently renewed and expanded their business with us. I met with the team there, and I was blown away watching them demo what they've built. They replaced their initial logging solution with Elastic, which has become an integral part of their security analytics, customer service and network operation infrastructure. I love it when several use cases are woven together and have a compounding effect on an organization. We've seen this at Samsung SDS, who closed new business with us in the quarter. I had the pleasure of meeting with them last week at our Elastic{ON} Tour event in Seoul. They used Elastic for security analytics and many other use cases such as enterprise search, app search, logging, infrastructure monitoring for many Samsung businesses within the Samsung group.

Another example of this is Australia's largest telecommunication company, Telstra, who renewed and expanded their business with us in the quarter. They've migrated from their previous search appliance to Elastic Site Search Service for search across their internal and external websites. They also use our software across several other teams to power threat detection, security analytics and network monitoring across high-profile customers. We live and breathe to be able to enable this kind of flexibility for our users.

Another important aspect in the security space is geo data analysis, for example, identifying the location of an IP address participating in a DDoS attack. That's a geo problem. You can imagine how being fast really matters in a scenario like that. This is true for other use cases as well. When you filter results based on location on a ride-sharing app like Uber, that's a geo problem, and fast results matter. If you're stuck waiting too long, you may just find a lift elsewhere.

So geo has always been core to what we do. It's a first principle feature of our products, and we've continuously invested in making it a fast experience in Elastic. How have we done that? A few years ago, we implemented a type of data structure called Bkd tree in Elasticsearch, and what it does is allow for 3 things: significantly faster indexing; faster queries; and more efficient storage for geo use cases. Up until today, it was only applied to what's called geo point analysis. Think dots on a map, latitudes and longitudes. As of 6.6 release this quarter, we have applied Bkd trees to geoshape analysis. So we move from analyzing dots to areas, which can be more involved because we're talking about the analysis of more complex shapes like bike trails, boating districts, state borders and bus and shipping lines.

We've heard from one geo customer that they're seeing nearly 30x performance improvements for this type of geo analysis in version 6.6 compared to 6.5, which I find very exciting. Making things faster is also a thing at Elastic. So you can start to see how these improvements are just one step we're taking towards geo potentially playing a bigger role in the Elastic solution space, and we're very excited about that.

Before I close, I'd like to talk about developments with our APM solution. I'm super excited about Elastic APM now being available on our Elasticsearch Service and Elastic Cloud Enterprise, aka ECE 2.1. I can't wait for our users to get out there and try it. The team really has done a wonderful job.

We also released a major feature in Elastic APM called distributed tracing. This feature gives users the ability to observe complex interactions happening between their services and applications. If a web page is loading slowly, any number of things could be causing it. Is it the browser? The server? The database? A third-party system? Maybe a back-end API? With distributed tracing, users get full visibility into all of this and can quickly pinpoint what's wrong. This is a key capability as more and more applications move towards distributed and micro services-based architectures.

I'm also proud to say that all of our APM agents are compatible with OpenTracing, which is a vendor-neutral way to introduce distributed tracings into applications. OpenTracing is a project of the Cloud Native Computing Foundation. I share this because we formally became members of this foundation in the quarter. This is meaningful to us for a few reasons: it strengthens our commitment to open standards; it demonstrates our capabilities with technologies like Kubernetes and Prometheus, we even reviewed Helm Charts in Elastic in the quarter; lastly, it shows our investments in cloud native technology stacks and the observability space. Continuing to invest here remains important to us, so much so that we've organized some of our engineering teams around it.

As you can see, Q3 was a strong quarter for us. Our users and customers continue to embrace our search products that address an expanding set of use cases. Given the incredible customer and user demand that we are seeing, we are further accelerating investments across all parts of the business. We look forward to our continued momentum as we close out FY '19.

And with that, I'd like to say thank you all very much, and I'll hand it over to Janesh to talk about our financial results from the quarter in detail. Janesh?

Thanks, Shay, and thanks again to everyone for joining us. We executed well during our first full quarter as a public company, and I'd like to express my appreciation to all our employees and their families for the hard work that went into this. I'll first go through our results for the third quarter of fiscal 2019 before discussing our outlook.

We performed well across all dimensions in the third quarter: robust customer growth, a high net expansion rate, and strong billings and revenue growth across all geographies. As Shay mentioned, total revenue for the third quarter was $70.8 million, growing 70% year-over-year. Subscription revenue totaled $64.6 million, an increase of 65% year-over-year, and comprised 91% of our total revenue. Within subscriptions, revenue from our SaaS products was also strong at $11.7 million, growing 67% year-over-year. While we remain very excited about the SaaS opportunity ahead of us, we remain agnostic to customers' preferences on how to purchase our subscriptions, whether SaaS or self-managed. Over time, we expect SaaS to gradually increase as a percentage of revenue based on customer adoption patterns.

Professional services revenue was $6.2 million, an increase of 136% over the same period last year. In the quarter, we saw strength in training and a healthy ramp in our consulting business. Professional services revenue is typically recognized at the point in time the services are delivered and can therefore fluctuate from quarter-to-quarter. We expect professional services will remain a small proportion of our overall revenue as our business grows.

In terms of geographic breakdown, given our roots in open source and our globally distributed model, we continue to enjoy strong growth outside of the United States. In Q3, 44% of our revenue came from outside the U.S. We saw broad-based strength across all geos, and in particular, EMEA. Our investments in the EMEA region have put us into position to scale in the coming years. We believe we have a rich market opportunity across the world, and remain dedicated to investing appropriately to capture that opportunity.

Moving on to calculated billings. We define calculated billings for any quarter as total revenue recognized in the quarter, plus the sequential increase in deferred revenue as presented on our statement of cash flows, less the sequential increase in unbilled accounts receivable. Calculated billings in Q3 was $79.8 million, an increase of 68% year-over-year. The mix of short-term deferred revenue was consistent with prior quarters at 92% of total deferred revenue.

At the end of Q3, remaining performance obligations totaled approximately $304 million, of which we expect to recognize 86% as revenue over the next 24 months. We were pleased with the growth of the remaining performance obligations, which signal continuing long-term customer commitment to our technology.

As we've said before, calculated billings can fluctuate from quarter-to-quarter based on the timing of renewals and billings duration for larger customers. Given this, an additional way to look at calculated billings growth is on a trailing 12-month basis, which provides a longer-term view of the business. Trailing 12 months calculated billings growth ending Q3 was 76%. This compares to 77% in Q2. We are very pleased with the calculated billings growth this quarter and the underlying demand that is driving our business. The strong growth in Q3 was driven by a broad array of growth vectors including new customer additions, new use cases at existing customers and larger deployments.

At the end of Q3, we had over 7,200 paying subscription customers compared to over 6,300 such customers at the end of Q2. We continue to see strong momentum with new customer additions. We also remain focused on growing the number of our larger customer accounts and ended the quarter with more than 380 customers with an annual contract value above $100,000 compared to more than 340 such customers at the end of Q2. Our existing customers continue to expand their relationships with us, reflecting increasing spend for existing use cases and adoption of new use cases.

In Q3, our net expansion rate remained over 130% for the ninth consecutive quarter. As a reminder, we view anything above 130% as being best-in-class, which is why we consider that a relevant benchmark.

Overall, we were pleased with the pace of customer additions and customer expansion. When viewed collectively, these metrics provide insight to our execution against the enormous market opportunity ahead of us.

Now turning to profitability which is non-GAAP. Gross profit in the third quarter was $52.6 million, representing gross margin of 74.3%. Total subscriptions gross margin was 81.1%, in line with the past couple of quarters. As planned, we continued to make investments in the SaaS business in Q3 and are tracking well relative to our expectations. In the near term, we will continue to invest in our SaaS business, and it will remain a modest headwind to gross margin overall.

Our professional services gross margin was 3.6% as we added further capacity in advance of revenue and as existing hires ramp to productivity. Since the professional services business is small, even relatively insignificant amounts can swing the gross margin in either direction. So we expect that the gross margin in professional services will fluctuate significantly from quarter to quarter.

Turning now to operating expenses. We remain focused on investing to drive topline growth. Overall, growth in spending was lower than planned, driven primarily by slower-than-expected headcount increases. The holiday period had a greater impact than we expected. To a lesser extent, there was some timing-related expense pushouts to Q4.

Sales and marketing expense for Q3 was $33.7 million, up 71% year-over-year, representing 48% of total revenue. While we expect to realize leverage in sales and marketing as we scale the business, our primary near-term focus remains adding sales capacity and expanding market coverage as we drive growth. We plan to further expand our sales team in order to grow in new markets as well as deepen our penetration where we already have sales coverage.

R&D expense in Q3 was $21.1 million, up 55% year-over-year, representing 30% of total revenue. R&D remains a major investment area as we expand our innovation advantages. Shay earlier provided a number of examples where we continue to invest heavily in both existing and new products to address the growing number of use cases.

G&A expense was $9.6 million, up 45% year-over-year, representing 14% of total revenue. This includes costs associated with our global expansion and continuing to build the infrastructure to scale for the future.

Our operating loss in the quarter was $11.7 million, with an operating margin of negative 16.6%. Net loss per share in Q3 was $0.16, using 70.7 million basic and diluted shares outstanding. This compares to a net loss per share in Q3 of last year of $0.27.

Free cash flow was negative $9.9 million in Q3 compared to a negative $2.8 million in the same period a year ago. As a reminder, there are both seasonal and timing effects in cash flow. Free cash flow is seasonally stronger in the first half and weaker in the second half. There can also be some lumpiness to inflows and outflows in any particular quarter. While we don't formally guide to free cash flow, we expect Q4 to remain negative with a modest improvement in free cash flow margin for the full year FY '19 versus FY '18.

Turning to the balance sheet. We ended the third quarter with approximately $306 million in cash and cash equivalents. We remain comfortable with our cash position from an operating perspective.

Lastly, we ended the quarter with 1,247 employees, adding 118 people in the quarter across all functions. We are adding people across the board to support the growth of our business and expect headcount growth to accelerate in the fourth quarter as we further invest in our business.

Moving on to guidance. Before I provide the outlook for Q4 and the full year, let me share with you our investment philosophy. We expect to maintain a disciplined and thoughtful approach to drive future topline growth. Innovation remains a top priority for us, and we will continue to invest in R&D. At the same time, we will continue to invest in sales capacity and coverage globally. While we are still working through our FY '20 plans, given the significant market opportunity, we expect to further accelerate headcount-related investments. Some of these investments will be intended to secure growth in FY '20, while others will help drive growth only over the long term.

Unrelated to these investments, I'll also point out for modeling purposes that we expect FY '20 expenses to be weighted towards Q1, given timing of internal global corporate events as the company continues to scale. I plan to share additional details regarding our FY '20 outlook at our Q4 earnings call.

Turning specifically to the fourth quarter and the full year fiscal 2019. For the fourth quarter of fiscal 2019, we expect revenue in the range of $74 million to $76 million, representing a growth rate of 51% year-over-year at the midpoint. We expect non-GAAP operating margin in the range of minus 26% to minus 24%, and non-GAAP net loss per share in the range of $0.30 to $0.28 using approximately 74 million ordinary shares outstanding.

For the full year of fiscal 2019, we expect revenue in the range of $265 million to $267 million, representing a growth rate of 66% year-over-year at the midpoint. We expect non-GAAP operating margin in the range of minus 22% to minus 21%, and non-GAAP net loss per share in the range of $1.13 to $1.11, using approximately 56 million ordinary shares outstanding.

In closing, Q3 was an exceptional quarter. We delivered topline growth of 70%. I'm very excited about the strong results, and I look forward to sharing our progress with you at our next call.

With that, let's open it up for questions. Operator?

(Operator Instructions) Our first question comes from Mark Murphy of JPMorgan.

Shay, you mentioned that Groupon is replacing Splunk with Elastic. I think you also said Choice Hotels switched to Elastic and that Ricoh Group replaced a logging solution with Elastic. So I didn't catch who the other 2 incumbents might have been, and I'm not sure if you're disclosing that. But my question is whether you're seeing some kind of a sustainable trend change in that type of replacement activity as opposed to greenfield? Or were those just a few one-off scenarios that happened to occur this quarter?

Sure. I can take it. Mark, I would say that nothing fundamentally changed versus the previous quarter in terms of our growth in the context of the logging use case. We're both getting into what we would call greenfield use cases. The need to log never goes away, and new projects pop around whether within our own organizations or new organizations are happening. And as you know, we're one of the most popular open-source logging solutions out there. But at the same time, there's existing investments in logging solutions with other vendors that simply either age or becomes difficult to manage, and this is where we shine as well. And so we have this, I would say, [pincer-like] adoption model that just continues to perform the same way that we have historically.

Okay, great. And then for Janesh, how did the U.S. federal business fare during the shutdown? And would you say that all the agencies moved ahead as scheduled?

Mark, so broadly speaking, I think we're very pleased with the performance overall that we saw. The federal business for us didn't see a significant impact associated with the shutdown. Obviously, we are coming off the prior quarter, which was seasonally strongest because of the government's year-end. And in many instances, the agencies with which we deal weren't really impacted by the shutdown itself, so we didn't see any significant impact associated with that. It was the quarter that played out as expected for us in that space.

Our next question comes from Raimo Lenschow of Barclays.

Shay, you -- we talked earlier that a big part, 44% of your revenue, comes from international. Obviously, those markets seem a little bit more volatile given the macro, but also usually are kind of happier to adopt open source. Can you talk a little bit about what you're seeing internationally in terms of how customer behavior's evolving there?

Sure. Raimo, yes. So this again continues on our historic execution. It starts with the open source as a distribution model, which knows no boundaries or country lines, if you will. And that means that our adoption levels are pretty sustainable across the whole world, which we're very excited about. This also means that we, as a distributed company, have a footprint globally that is probably different compared to other companies in our realm, both of which we're very excited about. When it comes to whether Europe is slightly more volatile versus others, it's very hard to talk about macroeconomics and what would happen one way or another. What I am excited about is that open source tends to be -- open source in the free distribution model tends to actually benefit sometimes with a certain downturn in macroeconomics, and I think that can work to our advantage over the long term. And I will also say that when it comes to open-source adoption, we don't necessarily see Europe as something that outshines the U.S. or other regions when it comes to pure open-source adoptions. We just see it constantly happening across the whole world. I just came back from Korea, and we had an event there with 450 attendees and they're adopting our software quite happily as well.

Okay, perfect. And then can you talk a little bit about this, there's a good few software companies out there that use Elastic as a foundation and then just have kind of solutions on top of that, some of them you kind of acquired, especially I'm thinking about the APM space. Can you see how your relationships with these guys, with the guys that are trying to build on your platform, is evolving as you go and kind of grow yourselves into some of these product areas?

Sure. So -- I mean, our approach to customers, to companies that build on top of our product is to make sure that they're as successful as possible. Any usage of our product means that we succeed as a company. And again, it goes back to the fact that we make it very easy for users to go and build and innovate on top of our product. If we have -- if we happen to get into a certain market, the success in our market -- in this market versus others that have built on top of it, to us, it's -- the fundamental technology is our software, we're succeeding regardless. So in a philosophical way, we're, I would say, more on the Microsoft camp, if you will, when it comes to trying to make sure that we build the foundation that companies can innovate on top of it. And sometimes we'll get into it here and there. But still work very hard to make sure that our partners and our customers are successful in this space.

Our next question comes from John DiFucci of Jefferies.

I have a question for Shay and then a follow-up for Janesh. Shay, just thinking about the use cases -- and we talk to people in the field and we're trying to do more and more of that with you guys. We're hearing a lot of discussion and/or interest in the field on the security use case. I guess, how should we think about that? I know that's small for you now, and it's actually relatively new. But can we -- and you talked about it in the last call, too. But how do we think about that, how it's evolving now and how this sort of interest flows into sort of revenue, something that we can see?

Sure. John, so a few things there. I think that you can easily map our trajectory in the security space because it's not hard to take a logging use case, for example, and repurpose it to a security use case. At the end of the day, and I've mentioned it in the prepared remarks, a log can easily be used not only for operational use cases but also for security use cases. When I look at that space, the way that I think about it is it feels very similar to how we got adopted in the logging space about 3 or 4 years ago. And what people sometimes forget is that we actually started, then people compared us -- or users compared us to enterprise search solutions. And we actually went through an evolution of us being used in the context of logging. And I see that playing out now in the security space. Obviously, this is what we call the dance that we have with our users. We go to conferences, you go to RSA or our own conferences and we hear users adopting us, thought leaders, if you will, in the context of security use cases, and we listen to it as well and make sure that we have the right R&D investments internally to make sure that we build the right product in that context.

So we maybe just look back 4 years and think about security today the way that we thought about logging back then in enterprise search? Is that sort of how -- maybe we can translate it into at least a first case, a first look into how it transforms into revenue. Is that what you're saying?

Yes, that's the -- broadly speaking, and 4 years is a very long time frame, but this is how we think about it. I mean, I can tell you from a product perspective and a market perspective, we have people that know the security space. We're closing business today in the security space, and we're investing in both integrating and supporting all the companies that build on top of us in the security use cases. And we're also making sure that we're building the right level of product features in order to support more intimately and natively the security use case. So yes, I would say that what you just mentioned is correct.

Okay, great. Because we were -- we continue to hear a lot about it. And Janesh, listen, the numbers this quarter sort of speak for themselves. And I realize that license is a small portion of your revenue, and Anthony has been really good about making sure we understand that. But it was almost like $4 million greater than we modeled it. Was there any change to the average weighted duration of your contracts in this quarter that might have had a modest positive accounting effect on that line item?

No. Nothing that comes to mind, John. There was nothing fundamentally different in terms of the assumptions or in terms of the broader contract durations. For us, contract durations are -- have been roughly 1.5 years overall, and we are pleased with the duration in the quarter. It was consistent with what we've seen in the past. It can vary a little bit across quarters, to your point, but I think this quarter was straight down the middle of the fairway. And those long contract durations really are a reflection, in our minds, of the commitments that customers are making on the technology as they continue to use us in mission-critical environments. So we emphasize long-term contracts, and it's generally a good thing, but we didn't see any significant shifts this quarter.

Our next question comes from Heather Bellini of Goldman Sachs.

I've got a handful of questions. Some of these have been touched on in different ways, but figured if we could try and get a little bit more information. I guess, Shay, if you could just share with us, because the APM examples that you gave are really interesting. And I was just wondering if you could help us, what is it about your APM products that makes it easier to manage with Elastic over the incumbents? If you could kind of tell us what the key selling points are? And is there a way for you to share with us what percentage of deployments you think are using you for APM? And kind of how that penetration rate has been changing? And then the other question I wanted to ask a little bit about is with MongoDB right now, there's a lot of debate in the market about AWS being a competitor given they launched DocumentDB. This doesn't come up as much with you, but just in case it starts, I was just wondering if you can remind people of what you've heard from customers about why they choose to stay on Elastic versus kind of exploring the AWS-Elastic offering. If you could share with us a little detail there, if there's anything new from customers that you've heard.

Yes, sure. Heather, I'll start with the first one. Our goal with the APM effort is twofold at this point. The first one is to make sure that users that have invested with us, when it comes to the logging and metrics use cases, get additional value and insight from our product when it comes to adding APM data into our system. We truly believe that by being able to combine APM metrics and logging in a single system, you get a level of visibility into your infrastructure that is -- has just a compounding effect versus an additive effect. So I'm excited about that. And in my prepared remarks, you heard me mention that we're just users. And I go and talk in conferences, they just get excited about hearing that we now provide APM and being able to add that to their existing investments with our products. And that's also, by the way, our go-to-market efforts now. We don't go in and -- into what we call our greenfield customers and say, "Adopt us for APM directly." But we want to make sure that, to a degree, every single logging and metrics user will start to use us for APM. And that's more of a -- of the road ahead versus what's happening now across all the customer base. When it comes to the value that they're getting, I mentioned the visibility that they have across the board but that's also the fact that we're investing in APM as a product on its own, which I mentioned, and that's -- that both includes, for example, building agents for all the popular programming languages as well as investing in higher-level functions like distributor tracing. When it comes to actually competing, for example, head-to-head with other -- a pure APM vendors, that's not something that we do now. And -- but again, I think that our unique position and what I hear from our users is the ability to combine these 3 efforts -- 3 segments together. And to be honest, at one point in the future, we believe that these 3 aspects will just become the same data sets of the same use case, and we're positioning our company to be in that level. When it comes to the AWS question, if you remember, we mentioned it as well, we're set on a path where we actually double down on open. We took our commercial features and opened them and folded them into our products, and creating a free tier that only we can provide as part of our SaaS offering and self-managed offering. And that free tier basically allows users to still use our features for free but still allow us to be the sole provider of these features. As I mentioned in my prepared remarks, this is getting bigger and bigger and bigger. And just a reminder, we did it for the first time only in 6 -- slightly more than 6 months ago when we released it. But you can see already the trajectory of it when it comes to Canvas as an example as well as frozen indices, which significantly reduces the cost versus other people that use Amazon Elasticsearch open source. And even all the way up to APM. So APM is also a free proprietary features, at least part of it, and that's something that we just launched on our SaaS service as well. So that's the path that we've taken forward. It's slightly different than other companies. We're using our advantage with the open-source distribution model to start to drive more features that we will be the only provider and vendor that can provide it.

Our next question comes from Kash Rangan of Bank of America Merrill Lynch.

Two questions. One is, Shay, when you look at the security logging use case market, is it a view that with Elastic doing a good job that there is a -- you're broadening the market, the market itself is opening up more and adoption rates are picking up? Or is it that the replacement aspect is becoming a larger part of your story? And also, Janesh, one question for you. Looks like the number of new subscription customers you added seems to be growing pretty significantly. It's almost 2x from where it was a couple of quarters back. Is this likely to be a leading indicator of business? And also, if you can just finally touch upon hiring trends, and what rate you expect to grow your sales capacity the next fiscal year. That's it for me. I know it's a lot.

Yes, thanks, Kash, I'll take the first one. I will mention it again in the context of logging and how we get adopted in the context of the security space. So we see 2 types of adoptions. The first one is new companies that try to build a sim solution or try to build a way to be able to have visibility into their security events. These type of new companies, startups, all the way from start-ups to new projects within organizations, tends to favor more adoption of open source and tools, and they are -- tend to be the thought leaders, if you will, in this space. And we see the adoption happening in our -- in the context of our products. Yes, we don't yet have the end-to-end security solution or sim solution out there. We're working towards it. But these are people that value the speed and all the core capabilities that we have, and they're joining us basically on this journey, the same way that they joined us 4 years ago when it came to the logging journey. At the same time, on the other side of the equation are companies that made investments in existing sim solution, and they just start to, to be honest, fall over, and try to put more and more data into them. And they struggle when it comes to being able to provide relevant results in a timely manner. And as you can imagine when it comes to the security space, this is critical to provide. And so these companies, those are large companies, are willing to again roll up their sleeves until we have an end-to-end solution in the sim space and replace or augment some of their existing sim solutions just because of the criticality of that in their business. These are the 2 ends of the spectrum, if you will, that we see mostly in the context of the security space. And obviously, we're working towards broadening that all the way to the whole spectrum.

And Kash, this is Janesh. In terms of the other couple of questions. On the total customers, we are obviously very pleased with the way the team executed here in Q3 getting up to more than 7,200 total subscription customers. And if I think about that just in absolute terms, you're right, it's an increase. The other way to think about it is in terms of year-over-year growth. And it's, broadly speaking, consistent with the pace that we see in the business overall. It's a little bit tricky to think of that as a leading indicator just because there is a long tail and the future growth really depends on adoption patterns and growth patterns that you then see within that broad base. So I'd be a little bit careful about doing that. And then finally in terms of the investment profile for next year, we are still working through our plans for fiscal '20, so I don't have a number that I can share with you in terms of how and where we will specifically invest. But I can tell you that based on what we see out there as the market opportunity, which is really the same as we perceived it 90 days or so ago, we are continuing to look at investments really across the board in all sales functions, both deepening coverage in territories where we already have presence as well as expanding into newer territories.

(Operator Instructions) Our next question comes from Matt Hedberg of RBC Capital Markets.

Shay, in your prepared remarks, you talked about working to broaden the consumption to more users. And clearly, you highlighted a lot of the investments you're making from a technology perspective to enable this. But I guess I'm wondering, can you give us a little bit more detail and color on some of the other things that you're doing to drive accelerated consumption? And I guess, maybe specifically, can geo size play a bigger role in terms of expanded distributions? What sort of investments would be required there?

Sure. So I think that I can use one example and then at least tell how we think about it. So I'd like to think about Canvas as a way to broaden the consumption of our software. I'll reiterate the story of Canvas. We built Kibana as a way to visualize data in Elasticsearch, and people have been very successful with building dashboards that covered it all the way from logging to security to metrics to other use cases. But what we saw is that these dashboards started to pop up, not only in front of operations people but also in the entrance of companies or at the office of CSOs and CIOs and CTOs. And when it comes to that, we actually believe that for them to be proud of data that is the fiber of their business, we need to provide them with a tool that they'll stick to and actually be happy when they show it around. And so that's why we created Canvas, because it allows basically to create -- partially it allows to create this level of expressability of data and something that you're proud to put at the entrance of your company hallway -- on the hallways or a CSO when they show off their security implementation. So that's one example where the users are actually either first-order users, which is the C-level executives, or second-order users, which are the people that walk the hallways and see the -- how proud a company is of their data. So that's at least one aspect of it.

And then maybe just to touch on the GSI question more specifically. We have partnerships with a number of GSIs, Matt, as we've touched on before, many of those in the early days. And the thing to keep in mind with the large GSIs is that they as well as other partners, will go where customers eventually lead them. So we focus on really building those deeper and longer relationships. And then as the business continues to grow and scale, I think that's when the -- some of the larger GSIs will be in a position to better invest and build practices around Elastic.

That -- no, that's super helpful, Janesh. And then maybe one other one for you. In your prepared remarks, you talked about trailing 12-month billings as the way that we should look at calculated billings, and I think that's certainly helpful to keep that in mind. I also think in the past you've talked about seasonality of billings being the strongest in Q2 and Q4. Realizing that this coming Q4 is a difficult challenge, is there anything else we should consider when we think about billings -- billing seasonality this Q4?

When I look ahead at the number, obviously Q4 seasonally is our strongest quarter, and a significant portion of the business for the full year gets done in Q4. So I don't see anything significantly different at this point in time. But what we have done is reflect the overall performance that we do anticipate in the guidance that we provided.

Our next question comes from Tyler Radke of Citi.

I wanted to ask you about the SaaS or cloud business. It looked like this quarter, the growth rate decelerated further, kind of, and is now growing in line with overall -- the overall business or overall subscription. I guess I would have been a little bit surprised to see that, like especially in -- a few months ago. Can you just talk about what's going on in that business? I know you did somewhat recently launch the service on Azure. So I just would have thought maybe there's some pent-up demand there.

Yes, I can take that. So first of all, we announced last quarter that we've released a major feature in our SaaS service, basically the ability to deploy what we call hot-warm architecture. That architecture enables more use cases to be deployed on our SaaS service, specifically in the logging and metric use cases. But at the same time, we took the opportunity because it allows to deploy in use cases. We took the opportunity to reflect back some savings that we have in infrastructure costs back to our users. Like any good SaaS provider out there, that is expected of us. That also means that, financially, users can onboard more easily more use cases to it. As a result of it, obviously, the pricing changes. And this aspect, you should see some changes when it comes to year-over-year growth, well, I will say, thanks to that. At the same time, we're just seeing it playing out now. We're hearing users getting excited about the fact that they can onboard more use cases into our SaaS service but that takes time to migrate and move, and that's how we see it playing out. So I'll be honest, I think that our SaaS business is relatively at the same pace. We're happy with how it grows. And as Janesh mentioned in his prepared remarks, kind of like building a company that is agnostic to where the user will end up deploying their data as long as they use and engage with us as a company.

Great, helpful. And maybe a follow-up for Janesh. You've talked about how hiring was a little bit below plan. Obviously, the OpEx growth came in a lot lower than what you saw in Q3. Can you help understand -- are you seeing any challenges as it relates to the hiring environment? I know you talked about underhiring about a year ago, and obviously, it's a difficult job market out there for -- especially attracting talented developers. But maybe just some observations on what you're seeing out there, and what gives you confidence you can resolve that here in Q4.

Great question, Tyler. So in terms of what we see out there, it's really not any different. The -- I wouldn't characterize it as hiring challenges by any yardstick. I think the community of users that we have out there continues to be very excited about Elastic. It's relatively the same in terms of hirability to attract talent both on the sales side as well as on the development side and clearly in the G&A functions as well. As I mentioned earlier, the reason hiring was a little bit slower was we just didn't anticipate the effect of the holidays and how much of an impact that would have. And then there was also a little bit of other expense pushouts that I referenced that caused OpEx to be a bit lower. So we feel confident that we can continue to attract the right talent, and there's still a pretty healthy pipeline of candidates and people that join Elastic every single month and every single week of the month.

This concludes our question-and-answer session. I would like to turn the conference back over to Shay Banon for any closing remarks.

Yes. Thank you. Thank you all for joining the call. As we mentioned, Q3 was a strong quarter for us. We look forward to executing against our plan for the rest of the year and updating you next quarter. Thank you all very much.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.