We think the shares of search company Elastic (ESTC) are still a long-term buy because the company has a number of fairly unique characteristics that provide it with a strong competitive position, rapidly increasing use cases and addressable market.
However, given the steep valuation, investors need to be able to stomach the short-term sentiment swings and a considerable hurdle is also in front when the lock-up period from the IPO expires.
Elastic enjoyed a highly successful IPO early October at $36 a share and the shares have more than doubled:
Elastic is one of these SaaS companies whose business model seems to have endless opportunities to lock in users and expand their usage. We have used our generic business model for these kinds of companies consisting of the following elements:
1) Get a foot into the door with a killer app, something at which the company is good at that fulfills a real business need. Indeed, Elastic’s killer app is search, from the Q2CC:
at the heart of the stack is the Elasticsearch. It’s what stores, searchers and analyzes data, structured or unstructured, and it’s what everything gets built around… Elastic is a search company. We believe search is foundational for a wide variety of experiences and use cases. You may not realize it but you probably use Elastic every day. When you catch a ride on Uber, we are the engine that matches the driver with you; when you look for groceries on Instacart, Elastic provides you with relevant results and recommendations; when someone swipes left or right on Tinder, Elastic gives powers finding a match you might like and who might like them back.
Management argues that speed, scale and relevance set them aside from the competition, but we would argue that it’s got more to do with the open source nature and malleability, enlisting users to create new use cases (see below).
2) Have a nice side business called services where you help customers understand the product, show what it can do for them and train them and help in installation, configuration, etc.
The company generated $5.1M in revenues from professional services in Q2, growing at 128% y/y but still generating negative (4.9%) gross margins.
3) Open a partner channel, that is, build a community. The company is doing that because it’s open source, see below.
4) Use the recurring (subscription) revenues to build out sales and R&D. The company has more than 6,300 subscribers (mostly on-premise, but also in the cloud), 800 more than at the end of Q1.
92% of their revenue (in Q2) comes from subscriptions and the company has 340 customers with an ACV of $100K+ (40 more than at the end of Q1). Needless to say, the company is investing lots in R&D (as well as S&M).
R&D spending is 32% of revenue and grew 88% y/y (S&A is even bigger at 50% of revenue, growing 97% y/y whilst G&A is 14% of revenue and growing 80% y/y in Q2).
5) Use R&D to build additional functionality and/or verticals, modules that can be used to up-sell (‘land and expand’). The company is adding functionality rapidly (Q3CC):
at the heart of the stack is the Elasticsearch. It’s what stores, searchers and analyzes data, structured or unstructured, and it’s what everything gets built around. Beats and Logstash are ways to ingest data into Elasticsearch from many sources, and Kibana is how you visualize data in Elasticsearch. We also build solutions that are vertically integrated into our stack. They include app search, site search and enterprise search; logging, metrics and APM; business analytics and security analytics.
And this is by no means complete but it’s clearly working, given the company’s net retention rate of over 130% for the last 8 quarters.
6) M&A might be used for the same purpose as R&D, to acquire new capabilities to up-sell. Indeed, the company acquired a host of new capabilities through M&A, for instance:
7) Open up the platform for third party/customer apps and take a share of the cut or use it to solidify the platform position and value. As we explain below in a little more detail, the company has been open source from the start, enlisting partners to generate new use cases.
8) Grow revenues in order to achieve operational leverage. However, the company is not there yet, they’re still trying to grow as fast as they can.
9) Ultimately, earn enough free cash flow to deleverage (where applicable) or allowing the company to buy back the shares that are issued as stock-based compensation and/or pay dividends.
The company is still using cash, although less than we feared (free cash flow was just a negative $1.4M in Q2 and even positive $3.4M a year ago) and they have plenty of that left from the IPO ($318.6M, of which $264 was from the IPO).
But cash flows are seasonal, strong in H1 due to strong collections, and weaker in H2. In fiscal 2018, their free cash flow margin was a negative 15% but they do see a gradual improvement of this.
So all these elements are more or less present in the business model of Elastic, but there are three things that seem to make Elastic stick out even more as a company:
- Its core technology is almost a general purpose technology, generating widespread applicability.
- A unique business model.
- Endless opportunities to shape and add.
The uniqueness of the business model is that much of it is open source. This brings four important advantages:
- It constitutes a broad funnel with which to harvest customers.
- It produces endless malleability.
- It enlists a community of users to add to these, building ever more use cases on top of the existing stack.
- All this results in customized solutions that are very sticky.
The malleability not only comes from the business model, but also of course from the fact that search is a rather generic technology. In any case, malleability has been designed in. From the Q2CC:
ever since we created Elasticsearch and then when we formed the Company, our goal was to try to build a product suite that allows for very easily add many different type of data sources on top of Elastic, whether it’s through the ease of use of the API-driven development, whether it’s through the ease of use of creating visualization for many different use cases. Whenever we develop a feature, we think about it from a pure search perspective and then we’re always curious about the fact how will that end up applying to many different use cases, the new use cases down the road.
But since much of it is open source, it allows customers to do the same and generate even way more use cases, CEO Bannon (Q2CC, our emphasis):
I was 100% sure that search applies to many use cases, I just didn’t know which ones. But, then, someone decided to put a log message in Elasticsearch and they decided to put the log message in Elasticsearch but not in any type of the enterprise search products out there. They decided to put their log message in Elasticsearch and not in any of the NoSQL solutions out there, they decided to take a log message and put it in Elasticsearch and not into Hadoop vendors out, especially in the early days. And that speaks to the fact that we’re building products that allow to people to innovate and imagine what could happen when they put different types of data into Elastic and then see the results of it.
They go around customers and see new use cases and then they go back and build that in a more general purpose way (Q3CC):
But, one of the things that’s really made us successful is following the community and evolving the stack and the features in the stack, based on what the community’s and users’ needs are.
An example of this is Canvas, which came out when they saw Kibana dashboard’s permanent display in the Brazilian Ministry of Health and thought that was a cool thing to do (Q2CC):
Canvas is a way to take data that exists in Elasticsearch and expose it to a whole new audience that we didn’t necessarily see. And I’ll admit like, this is something that we saw happening and then we ended up developing it as a result of it. So, we’ve developed Canvas to try to expand these — the data that exists and to other audiences. So those are — to be honest, because we’re an open company and an open source companies, what we’re doing and how we’re moving forward and the investments that we do is out there.
That is, the company still invests substantially in R&D and they don’t see open source as a way to outsource R&D, and the company’s paid for services are still unique (Q2CC):
we are the only offering that provides features like Canvas, roll-ups, logs and infrastructure UI and many others that I mentioned earlier, no one else does. This is also true for the custom topologies feature, I just talked about. We are the only hosted Elasticsearch offering that provides this flexibility.
Open source also serves as a funnel to harvest clients as these are often organizations that were already experimenting with the open source stuff anyway (Q2CC):
Our goal is to engage with the customers at the point where they’re already using us. We don’t necessarily want to engage with customers when they don’t — haven’t used us yet. That’s the whole point of open source and free distribution model… Because users deploy our software long before they engage with the sales reps, sales engagements often start with warm leads and in many instances there is existing executive mindshare. All this leads to efficient sales cycles.
It’s difficult to consider search without machine learning to improve the underlying algorithms. While the company developed and acquired its own machine learning solutions, they display the same flexibility and openness to interfacing with other libraries (Q2CC):
We’ve built a foundation where you can store your data and execute search queries and search algorithms on top of that data, as I mentioned in an extremely fast manner. Our stack has always been extremely open. So, if you go and for example, see what the community has built on top of our stack, you would see integrations with TensorFlow, integration with R, [ph] integration with Pandas other very popular machine learning libraries and obviously we’re help the committee to drive this level of innovation… when data is stored in Elasticsearch, you can go and run almost any type Google machine learning algorithm out there on top of the data in Elasticsearch; that’s very easy and very simple integration. So, all the innovation that Google does in machine learning, immediately applies to the benefit if someone ends up putting data in Elasticsearch itself… But also, we are developing our own set of machine learning algorithms and built-in features that are integrated directly into the stack.
The latter is supposedly more simple to use without having to involve data scientists.
- Q3 revenue will grow 56% at midpoint to $64M-$66M.
- Non-GAAP operating margin will be negative 30%-28%.
- Non-GAAP EPS will be negative $0.32-$0.30 (on 71M shares).
Fiscal year 2019:
- Revenue growing 60% at midpoint to $254M-$258M.
- Non-GAAP operating margin negative 26%-25%.
- Non-GAAP EPS negative $1.35-$1.30.
Inevitably, for such a general use technology with an exploding amount of use cases the valuation is very steep. It has a market cap of $5.7B, cash holdings of $318M so an EV of $5.4B and guided revenues of $256M so the shares trade at 21x sales.
We think the chances are pretty good the shares are a solid buy for the long run. However, given the rather stratospheric valuation, potential short-term swings in sentiment could lead to considerable losses in the short term, especially after the end of the lock-up period next April.
Yet we think the uniqueness of the company, combined with exploding use cases and addressable markets, the company has a very long runway in front of it and the longer they can grow, the more ingrained their solutions become.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.