Snowflake — a fast-growing data cloud IPO breakdown analysis
Network effects, also known as Metcalf’s Law, occur when the value of a product or service to a customer increases by the use of the product by the next customer (more on that from a VC perspective here and from the FTC view on competition here). As the network continuously add new nodes, customers obtain exponential more value by having access to new opportunities and therefore will become less willing to drop out.
As a result, in industries that exhibit strong network effects incumbents become stronger as the network matures, and we can identify a ‘winners-take-all’ type of market share once the majority of the network adopts one platform. And although it’s easier to detect strong network effects in consumer business (think of Facebook for social media, LinkedIn for ‘professional social media’ and Microsoft Windows for Operating Systems), there are powerful network effects in the enterprise.
In these markets, B2B companies that conquer a segment become the de-facto platform for a defined objective (e.g. Slack for enterprise communication, Jira for Project Management tracking and Shopify for the direct-to-consumer e-commerce ecosystem) with very high switching costs associated (At Slack direct listing, 80% of paying customers continue to use the platform after 5 years), and therefore an attractive lifetime value (LTV) profile.
And Snowflake, the 8-year old San Mateo-based data cloud company that filled its prospectus to become a public company, is the next high-profile tech unicorn to prove the value of network effects in the cloud.
SNOWFLAKE BUSINESS OVERVIEW
The database industry has evolved tremendously since the days that Katherine Johnson, Dorothy Vaughan and Mary Jackson uses clunky IBM mainframes to calculate the trajectory of the Apollo mission.
Now, organizations have on average developed more than 450 custom applications to run their processes, each structured with a unique set of architecture requirements to solve for a unique set of business requirements — some are run on cloud environments and others are still on-prem, most often with a unique database. With that complexity, it is common for organizations not to have a unique database with information from your business, employees, suppliers and customers.
In this context, Snowflake was built to become a cloud data platform that sits on top of the three largest public clouds (AWS, Microsoft Azure and Google Cloud Platform) in more than 20 different locations across the world to offer the ability for enterprises to organize their databases irrespective of data type (structured or semi-structured), scale and performance needs.
The general architecture is structured by three interconnected layers: storage, compute, and cloud services. The first is the ingesting machine that organizes the customer different databases in one location. The compute layer allows users to effectively use the entire data set. Lastly, the cloud services layer is used to optimize each use case’s performance requirements.
What is unique in the Snowflake offering is the scale, performance and the breath of the data cloud ecosystem where companies share their data with suppliers and other enterprises. For instance, Snowflake mapped the network of hundreds of companies from different sectors that shared information related to the COVID-19 pandemic (each bubble is a database shared by a business within the Snowflake ecosystem and each link is an integration between different databases) and it is exciting to appreciate how the complexity of the network has added value to organizations that wouldn’t otherwise have a seamlessly opportunity to have access to these datasets.
By extrapolating this example, we can infer that the Snowflake ecosystem has created several powerful (and more complex) data clouds demonstrating that customers have being benefiting by sharing their data with their stakeholders. It also represents one of the greatest examples of the stickiness of the platform.
Product demos are a great way to understand how a business is in fact selling their services. At Snowflake, there is a new 20-min product demo outlining some of the benefits of the Data Cloud. Interestingly throughout.
Over the last four quarters the company has grown its customer base by an average of 139% and more than double the number of customers in the most recent quarter compared to the prior year. It’s also worth mentioning that nearly 30% of the companies among the Fortune 500 are Snowflake customers representing a total of 26% of the company’s total revenue. More interesting is that the cohort of large customers — defined by Annual Contract Values (ACV) greater than $1 million — is growing faster than the total customer count. As of 2Q 2020 there are 56 customers with ACV greater than $1 million, a number 155% greater than same period of the last year.
Another important metric to analyze is the net retention rate, which is calculated as ratio of revenue by a cohort of customers in the second year divided by the same cohort’s revenue in the first year. Over the last 8 quarters, net retention rate has a mean of 180% and median of 176%. According to Redpoint Ventures, the top quartile of net retention rate for SaaS companies is 124%, meaning that Snowflake significantly outperform its broad SaaS peers. This means the company is quickly adding new logos, especially large accounts, and customers are constantly spending more on the Snowflake platform.
Another way to measure the success of the platform is to look at the quotient between LTV (lifetime value) and CAC (customer acquisition cost). And there is no one better than the team at A16z to explain the importance of this metric.
But before crunching the numbers, these are the assumptions used in the calculation:
- For CAC, I divided Sales & Marketing spent in a quarter by the net new customer count in the same period.
- For LTV, the average lifetime is five (the amortized sales commissions period that Snowflake determined) and the average contract is the revenue per customer at a quarter x 4 x net retention rate to capture future contract expansions
With that being said, LTV/CAC always above 3.5x over the last seven quarters and an average of 5.5x. The rule of thumb for SaaS businesses is that this ratio should not be below 3 meaning that the business is growing at a healthy ratio, capturing market share and is poised to generate future positive cash flows.
The next segment is dedicated to analyzing the leadership of the company, which is kind of unique for venture capital-backed companies in the sense that co-founders and investors have decided not to have a founder in the CEO since its inception.
There are studies that demonstrate that founder-led companies in the S&P are more innovative, create more patents and spend more in research and development to keep investing in future growth of firms while maintain profitable growth over long periods of time compared to professional-CEO firms.
Another line of research shows the importance of diversity in leadership roles to drive innovation and performance, with nearly 10% higher EBIT margins for companies with diverse management teams than that of companies with below-average diversity.
In the case of Snowflake, the company is below-average in terms of gender balance at the top of the firm with only two women (who are the chairs of the Marketing and HR functions) sitting at the 11-person leadership team.
Career-wise, the management team is heavily tech-experienced, with 3 former ServiceNow executives (including CEO and CFO), three xGooglers and also EMC/Dell former executives and both co-founders who worked at Oracle prior to founding the company.
One of the founders, Benoit, have appeared in this very interesting panel. Good to hear about his background and his views of the company. I could not find relevant videos of the other co-founder, Thierry, except for this quick PR video and this interview he gave for a French TV channel.
Speaking of founders, what sets Snowflake apart from the traditional venture capital-backed companies is the lack of a founder in the Chairman/CEO position. From inception the two technical founders chose to partner with outsiders. First with Mike Speiser, managing partner at Sutter Hill and board member, followed by Bob Muglia, who led the company from 2014 to May of 2019 when current Frank Slootman was announced the CEO.
The board has 9 seats, three of which are women (still not enough, but a good sign that companies are paying attention to gender balance in their boards). There is a good balance between investors (two), the CEO and independent members and I suspect the co-founders have chosen the CEO to be their representative in the board in a good sign that they understand their roles in developing the product and the business by letting an outsider run the board, a task that can be time consuming and complex.
FINANCIALS & COMPARABLES
Analyses of tech companies’ income statements are focused on net losses, driving cheap conclusions that a company is a loss-making machine. It is also very common to read in tech forums how high-growth tech startups sell ‘$10-dollar bills for $8’. Instead, the idea here is to demonstrate why and how can a SaaS generate very attractive future cash flows.
But first, it is important to recognize the short-term and long-term implications of the mismatch of revenues and costs in the typical SaaS business. These businesses have to incur elevated costs to build their products upfront before selling to customers. Once they sign a contract with customers (typically ranging from 12–36 months), for accounting reasons, these businesses recognize the revenue as the service is consumed. As a result, SaaS businesses will typically have higher upfront costs, which declines meaningfully overtime as revenues are generated.
And this is the case with Snowflake.
Over the last eight quarters the company has being managing its P&L consistently by sequentially reducing the percentage of its costs as a percentage of revenue. In 3Q18, costs were 236% of revenues with Sales & Marketing (S&M) alone representing 113% of total revenues. By comparison, in the most recent quarter, expenses reduced to represent 158% of revenue, and S&M as a percentage of revenues of 70%.
On a year on year comparison, revenues have grown at higher levels than any of the expenses over the last three quarters. The following chart shows how revenues are consistently doubling over the last four quarters whereas cost have been growing at stable — and lower — levels. This is a clear signal that the company will continue to become marginally more profitable, as revenues are growing at a larger base and at a faster pace.
Comparing Snowflake’s LTM (last twelve month) revenue and last quarter’s YoY revenue growth against a basket of 17 high-growth SaaS at their respective IPO grants a unique perspective in how strong their business has fared. It is the only in the cohort with $400m+ in revenue and +100% in YoY revenue growth. Zoom and CrowdStrike were the only other companies with +100% revenue growth at IPO, but at significant lower revenue levels (Zoom was at $330m and CrowdStrike at $250m).
On a revenue basis, Snowflake is the unicorn among the unicorns growing at faster rates and at a larger revenue base than all high-growth SaaS that IPOed over the last 18 months while competing against the three cloud giants that offer similar services. There is a lot to appreciate in how the leadership team has developed Snowflake and it will be interesting to understand how the network effects within their Data Cloud environment will evolve and how the market will value this company.
Don’t use this as investment advice: it is not.