In Part 1 and Part 2, I laid out a framework for Snowflake’s “data ecosystem” business model. I described it as +3 complementary platform business models with +4 network effects. And I think that explains most of what the company is doing right now.
However, I also recently listened to an interview with CEO Frank Slootman and it was interesting to hear his description of the business. It’s pretty consistent with my summary but uses somewhat different languages. Plus I think a couple of graphics would help.
But what was really useful was the three big benefits he cited for Snowflake’s users.
The first was how ease of use enables data transformation.
Slootman said Snowflake is mostly about getting companies to become data-driven for first time. By making data warehousing and analysis easy to implement, Snowflake is enabling companies to finally move away from slow decision-making and anecdotal analysis. And that is really how most companies operate. They do some tracking and reports. They try to do some analysis. And then they make decisions. But it is pretty fragmented and often anecdotal. It relies heavily on management judgement.
By enabling companies to become truly data-driven, it enables them to operate differently. They consolidate their information into one location and they began to run analysis on it all the time. Decision-making becomes much more data-driven.
- Slootman said companies usually begin by creating descriptive models from the data. What is really happening at the company?
- Then they go into predictive models, where the data is used to estimate what what will happen based on various decisions they could make.
- And then, eventually, they get to prescriptive models. This is when the data systems analyze, predict and then decide what to do.
In the process, data goes from informing management and staff to driving the enterprise and operations itself. It is about moving to a digital core, which I wrote about in Digital Operating Basics (see the Concept Library). Data and digital technologies become “the beating heart of business” (Frank’s phrase). This is really a complete change in how businesses have always operated, which is by people using reports and anecdotal observations. Frank said his customers are shocked when they learn they can implement the system and start running queries within a single day.
The second big benefit of Snowflake is the consolidation of a company’s data.
Data really is scattered across most organizations. At best, it is there are databases in various departments. More likely, it is just scattered across spreadsheets and reports. The various silos of data may be consolidated from time to time. But most of the data in an organization is known to few or simply lost.
Consolidation of a company’s data (i.e., getting it in one place) really does matters. It means every company can be a “big data” company. However, consolidation of data is more difficult that it sounds. The ease of data ingestion and unification is one of Snowflake’s biggest challenges and accomplishments. You can see the process of ingesting data as one of the key enabling capabilities for Snowflake’s platform.
A third benefit is Snowflake’s low latency.
Companies can start to run queries in real time based on incoming data. The speed of analysis can be a key differentiator key. Plus companies can do constant experiments.
I thought Frank’s comments were pretty interesting. And I didn’t see anything that contradicted my assessment. It’s also worth noting that he is making an absolute fortune running the company. Mostly based on options and performance.
That said, many of you have checked the share price and noticed that Snowflake has a whopping $100B market cap (10.21.21), which is around 100x current annual revenue. Not 100 times earnings. 100 times revenue.
That’s pretty amazing. Clearly, people are excited about the business model. And it’s growth potential.
And that brings me to the two big looming questions about Snowflake, which are its growth and competition.
Snowflake’s Growth Is the Big Unknown
I’ve been struggling with the growth question all week. How do you project out the growth of a company that is trying to digitally transform the operations of every major company? That is trying to become the beating heart of business.
I have yet to seen a growth model for the company that I believe. My approach has been to come at this from four angles.
- Look at early growth, growth hacks and virality.
- Project the growth of the core business.
- Estimate growth in adjacencies.
- Try to project a Total Addressable Market by use cases.
Slootman had some interesting comments about the company’s “data networking”. He said people are on Snowflake because of who else is on it. Corporate users naturally invite their customers, partners and suppliers to sign-up. Most companies want to connect and share the data within their supply chains. But this doesn’t happen much today because it is too difficult for most companies. Snowflake enables this type of “data networking”.
I think he is just talking about virality (see the Concept Library). Virality is a growth mechanism. It is when the usage of a product or service brings more users to the service. Given that Snowflake’s services are about connecting, consolidating and sharing, you can see how that happens. Virality drives growth, especially in the early stages of a company.
Slootman mixes this “data networking” idea in with the idea of “data network effects”. He also calls it a “data federation”. He talks about how the volume of data increases its value. And how combining and augmenting data also increases its value. And how the more granular the data, the more granular analysis. Basically, he is mixing together the idea of virality (which is a growth mechanism) with network effects. This is actually pretty common. Because network effects can actually have 3-4 different impacts on a company. Network effects can be a:
- Competitive Advantage
- Barrier to Entry
- Flywheel for Growth
This last effect (as a flywheel) is why companies with network effects can grow so fast in the early stages. As one company gets more volume, its service is inherently better. And in the early days of adoption of a new service, everyone jumps on that one company because they hear it is better. That helps it accelerate away. This early stage growth effect is sometimes called bandwagon effects.
And Snowflake is in the early stages of its growth right now. It is clearly benefiting from virality and a growth flywheel. That’s cool but I don’t think this is going to impact growth or value long-term. We should be ready for a slowing down of growth as these effects fade and it moves past this early stage.
Growth in the Core Platform is the Key
I did this week’s podcast on core vs. adjacency growth. Mostly because I was struggling to figure out Snowflake. I think that is a great way to think about Snowflake’s growth.
In that podcast, I listed 4 ways a company can both grow and adapt its core business. This is mostly from Chris Zook’s (Bain & Co) book Profit from the Core. He argues that most companies with sustainable growth have 1-2 strong core businesses. And these cores should be grown with focus – and also slowly adapted with changing times. This means consistently adding:
- New products and services.
- New customers, especially microsegments.
- New geographies.
- New businesses.
Chris was not talking about platform business models. But it’s still a good starting point for Snowflake. And we can see the company doing this. Platform business models are about creating interactions so the key numbers will users, interactions, data and cash flow. And Snowflake clearly is monetizing by volume of usage.
Growth in the Adjacencies is Also Really Important
In the podcast, I also listed six ways to grow adjacencies (also from Eric Zook). Looking at this list for Snowflake, it’s hard to find a direction they can’t grow in. They are sitting close to a long list of adjacency growth opportunities. They are flagged in the below list.
- New customer segments:
- Micro-segmentation of current segments**
- Unpenetrated segments**
- New segments
- New geographies
- Global expansion**
- Local expansion
- New channels
- New products
- New to world**
- Support services
- Next generation**
- New businesses
- New to world needs**
- New substitutes
- New models
- Capability adjacencies**
- New value chain steps
- Forward integration
- Backwards integration
- Sell capability to outside**
But the trick with adjacencies is to minimize the probability of failure. Here are the three factors that correlate with the success of such growth initiatives.
- Factor 1: Adjacency is tightly tied to a strong core.
- The economic distance is short. How much does it overlap?
- You need a strong core or a strong position in a channel, customer segment or product line in weaker core.
- Factor 2: It is an attractive adjacency market in terms of profit pools.
- Factor 3: There is a strong ability to capture economic leadership in that market. You need competitive advantage as an attacker and then an incumbent.
Snowflake is ridiculously strong in Factors 1 and 3, mostly because it is a complementary platform.
Can You Estimate the Total Addressable Market by Use Cases?
I’ve tried this and haven’t gotten very far.
Here is an estimate by Mario Gabriele in Demystifying Snowflake: The Biggest Software IPO in History (Sept 2020).
“ The International Data Corporation calculated that $88 billion in revenue was generated from global data storage in 2018, a figure expected to reach about $176 billion by 2023. The data warehousing market — Snowflake’s space — was considerably smaller but growing well. Estimated at $13 billion in 2018, the market is predicted to reach $30 billion by 2025, a 12% compound annual growth rate. As new devices and software programs increase the amount of data generated, demand for sophisticated warehousing solutions like Snowflake’s is likely to increase.”
Snowflake’s Other Big Question is its Scary Competitors
What about AWS, Google Cloud and Azure?
I wouldn’t want any of those companies as competitors. Yet, Snowflake has all three. And they are just peripheral competitors. Snowflake is right in the center of their core business.
- What happens when these companies bundle copies of Snowflake’s services with their other many services?
- What happens when start cross-selling with their direct sales teams?
- What happens they start subsidizing their prices?
I can think of 10 ways off the top of my head these companies could make life very difficult for Snowflake. Snowflake has had the advantage of first mover. But it has a window of time to get locked into its clients and to get its network effects going. Some thoughts by Mario on this:
“To date, the company has managed to prosper by being a sort of “Switzerland,” playing nicely with the Big Three’s offerings, meaning that customers don’t have to use just one. As more companies operate complex “multicloud” offerings, neutrality can be an advantage. But that may not always be possible. There are no identifiable barriers that prevent the Big Three from muscling in on Snowflake’s market share, and given their technical talent and funding, Snowflake is vulnerable.”
Also, keep in mind, Snowflake has other competitive advantages. Everyone likes to talk about network effects. But I almost always prefer switching costs. Snowflake is “critical enabler of digital transformation” for companies. How do they remove this once it is their digital core? When it is the center of their operations and management? The beating heart of their organization.
Snowflake also has economies of scale – in:
- R&D for new tech, new features / analysis and new integrations. Done by company and developers.
Snowflake is a cheetah.
Last Point: Snowflake’s Scalability Is Not as Clear as It Seems
Snowflake has a great growth story. But there is a problem. How scalable is their business?
There is the assumption that this is a digital company so it scales easily. And that is 100% true for software companies. But externally-focused AI companies are a different animal. They have to respond to lots of data types and qualities. And they must ingest and standardize that into the AI. This can require lots of ongoing labor to clean the data and ingest it. And the trained models can become obsolete and require ongoing retraining. The economics of externally focused AI companies are different than software companies. They look more like software plus services companies. They don’t necessarily have the same margins or easy scalability.
In the case of Snowflake, it depends on types of data being processed. The company specifically mentions it works with structured and semi-structured data. So it is not just letting companies stream in video feeds and lots of unstructured types of data.
I think Snowflake might be doing something quite clever. They have set the rules for their system and the users may be the ones doing the scrubbing, standardizing and tagging of data. They may have pushed the labor-intensive work onto their clients. This is similar to how TikTok and YouTube operate as video sharing websites. They don’t do any of the hard work of creating videos. They let users do that and then take the easy sharing component (which has zero marginal production costs). I think Snowflake is specifically limiting the data types and pushing the labor aspects to its users. They are keeping the most attractive economics for themselves.
That’s just a guess. But scalability will definitely become a problem as they ingest more data types. They are probably doing the easy stuff now. Their margins may be much less in the future.
That is my take. I think growth and competition are the two big questions. But I’m still struggling with them.
Cheers and thanks for reading, jeff
- Snowflake is Building 3 Complementary Platforms with 4 Network Effects (Pt 1 of 3) (Asia Tech Strategy – Daily Lesson / Update)
- Part 2: Snowflake is Building 3 Complementary Platforms with 4 Network Effects (Pt 2 of 3) (Asia Tech Strategy – Daily Lesson / Update)
- Core vs. Adjacency Growth in Digital Businesses (Asia Tech Strategy – Podcast 104)
- 3 Types of Network Effects (Asia Tech Strategy – Daily Lesson / Update)
From the Concept Library, concepts for this article are:
- Cloud Services
- Complementary Platforms
- Growth: Core vs. Adjacency
From the Company Library, companies for this article are: