Discover more from Kyle Poyar’s Growth Unhinged
Lessons learned scaling to $5B with Cockroach Labs’ CEO
Spencer Kimball on Cockroach Labs’ shift to the cloud and embrace of consumption-based pricing
A couple of weeks back I had a chance to hop across the pond, spending a few days at SaaStr and the Pavilion CRO Summit in London. There were incredible speakers across the three days, but I was perhaps most excited for my next guest: Spencer Kimball, CEO and co-founder of Cockroach Labs.
Cockroach Labs has had a fascinating journey. The company began as popular open source software (with 27k+ stars on GitHub), taking database technology used at Google and making it available for the masses. They now offer proprietary self-hosted, dedicated cloud, and – as of 2022 – serverless editions.
The company is clearly onto something promising, announcing a $278 million Series F in December 2021 along with news that Cockroach Labs’ cloud business was growing at 500% year-on-year (😲).
Cockroach Labs is part of a broader wave in infrastructure tooling that’s betting big on cloud and consumption-based offerings. We’ve seen similar moves from public companies like MongoDB, Elastic, Confluent, and HashiCorp.
Keep reading for Spencer’s insights on what matters when raising early funding rounds, how to scale an enterprise-grade technical product, and what he really thinks about consumption-based pricing in the current environment. The conversation has been lightly edited for clarity.
The TL;DR – Lessons learned scaling Cockroach Labs to $5B
Avoid working in a bubble in the early stages. The more people who you can get to give you feedback, the better.
The appetite for cloud services at the high end of the market has grown dramatically.
If you can just do the cloud service, just do the cloud service. Otherwise you’re building two separate businesses at once.
Selling cloud services takes longer than you expect, but time-to-value for the customer is faster. Start small, undersell, and expand quickly.
Ultimately you want to comp reps on consumption just like the customer is paying based on consumption.
PLG is a part of everyone’s strategy if you want to be successful, even if you sell to large enterprises. In Cockroach’s case, it’s about removing friction for other teams inside a larger account to try the product.
You have to collaborate with big cloud providers. Just don’t treat them as a monolith. Where you can win is finding where you're additive.
KP: What matters when you’re fundraising early rounds?
We'll start with what not to do. I see this somewhat commonly and I was guilty of it in some past incarnations, but you want to avoid working in a bubble in terms of your idea out of fear that someone else is going to pick it up and run with it.
People think ideas are very precious. That’s not the case in that early stage. Ultimately, they're not very valuable until execution has been applied. Honestly, if you look at your pitch deck several stages later, you’ll probably be vaguely embarrassed at how you described it, because it's just a raw idea and ideas change dramatically.
The more people who you can get to give you feedback, especially the ones that are in great positions to have good feedback like a venture capitalist, the better. The one corollary to that which is very challenging is if you even have a chance to talk to a competitor about your idea, I think that is actually something you should embrace. You get incredible feedback from somebody that's been thinking in your space as their primary occupation. VCs have a pattern match, but another founder that actually is working on something in the same space is going to challenge you pretty dramatically and evolve your thinking rapidly.
And, by the way, this advice extends to asking a venture capitalist to sign an NDA. Some people ask me if that's appropriate and my answer is absolutely not. Not if you want them to take you seriously.
But what should you do? In the early stage, I think you need to optimize for a VC that's going to be the best advisor possible. Think of it almost as who do I most want on my board?
And that's not always something that an early stage founder knows how to do. And so probably the best rule of thumb is who's got the most credibility here? What's their track record? What's the experience they have?
I found that the first investor sets a pattern that is extremely useful or sends a signal to the next investor that comes along. In fact, they often will just make the recommendation and someone will jump on it if they're credentialed enough so they can take the guesswork out in the next round. Early employees and early customers also pay a lot of attention to that signal.
And so if there's an opportunity, what are the things that you would trade on? What's the dilution, what's the valuation? Given the failure rate of startups, if you can get a very credible picker of winners early on to endorse you, that's something that I would trade a lot of dilution for.
KP: You started out open source and moved to the cloud later. How did you think about that decision in the early days and how has your thinking evolved?
Well, it's evolved a lot. When we started in 2015, the open core model was still ascendant and it was a perfectly reasonable thing for us to do.
Cockroach was already open source. And part of the reason for that is that I wouldn't personally use a closed source database. The cloud source ecosystem wasn't nearly as evolved in 2015, so that wasn't much of an option. Certainly the kinds of customers that wanted Cockroach in 2015 were not interested in consuming it as a cloud service. It was only small companies that were doing that and Cockroach was meant for bigger companies with scale or big operational resilience issues.
What we found is that since then the appetite for cloud services at the high end of the market has grown really dramatically. In 2015, very few of the bigger companies wanted to start there. I guess the perception of risk hadn't really caught up with the value.
And that's all changed partly because the risk has been addressed. How many years now since AWS first launched its relational database service (RDS)? It’s been a long time and there are a number of big companies that took the plunge early, whether they're born in cloud and scaled or they were just very aggressive about doing early adoption like Capital One. That's paved the way, plus the security and compliance sophistication has increased.
There's also the ROI. It has become clear that a cloud service provides a much better time-to-value and potentially a better total cost of ownership. I think that's true in most product categories now. And so what we're seeing today is that our biggest customers are either ready for cloud or they have a short-term roadmap to get to cloud. Short term in these bigger companies’ cases could be anywhere from one to five years, but typically somewhere in that one to three range.
It would be crazy not to build a cloud service today. My advice to everyone is if you can just do the cloud service, just do the cloud service. The difference between a company that sells a self-managed or self-hosted product and one that's selling a cloud service is vast.
It's different kinds of employees, product features, security integrations, pricing, and all the technology that's required to implement, for example, consumption based pricing. You don't need any of that in the self-hosted, but it's an extremely significant effort in the cloud service. It's different marketing, contract legals, different procurement processes. The support can be very different as well. If someone's running it themselves, you're actually supporting them in the operations. But none of that is true in the cloud service.
What you find is that if you do both, you have a lot of redundancy. It does feel like two different companies and ultimately everything's going to cloud services.
KP: Let’s unpack the challenges of going cloud-first. What were your top lessons learned in that journey?
Security and compliance takes longer than you expect, a lot longer. And at the enterprise segment, the buying process, the sales cycle is more substantial maybe even by months. These bigger enterprises take the security requirements very seriously. They have their own process and you have to fit into that and make them comfortable.
Things take longer fundamentally, but the time-to-value for the customer is faster. It amortizes that cost because you're going to seize faster expansion. And actually I think that's one of the really useful lessons with the cloud service. Start small, undersell and expand quickly.
Coming back to the problem of having both motions, it's an extremely different experience for your sellers. Ultimately you want them to be comped on consumption, just like the customer's paying based on consumption. That way it's less of this big supersized upfront deal that's supposed to maybe even be multi-year, and it's more you're buying credits and you want to undersell that and expand early.
The sellers have to get used to that. I think they make more money in the long run, but it's something to get their heads wrapped around. There's a transition in your sales force that has to happen. That is something to navigate, to put it mildly, because sellers are very focused on their comp and maximizing that on a quarter-by-quarter basis. Giving them a whole new process and less upfront and more in the long run, that's a hurdle to overcome.
KP: The interest in consumption-based or usage-based pricing seems to have cooled given the downturn. There are fears that consumption-based pricing isn’t as predictable and that there are a wave of spend optimizations. How do you think about consumption-based pricing right now?
This is one of the selling points of consumption-based pricing. In a downturn or during cost-cutting, the customers can and will cut their expenditures. They're like, well, we don't need to keep expanding so much, or we've been uncritical about how much data we're just throwing into this data warehouse. And then you get to the point where people are trying to save money and they're like, “Oh, we don't need to keep all this data. Let's take some of it out, let's archive some of it into cheaper storage.”
That actually is something that you want to lead with in your sales process. Again, I think that's where it's good to start small and really just try to deliver value for every bit of incremental credit burn. If you're always true to that, where the consumption is what the customer is actually driving value on their end for, you're less likely to see a dramatic reduction.
The exception, of course, is if your customer's business is seeing a dramatic reduction. In that case, you don't want to be overcharging them. Even in those multi-year deals if you don’t have consumption-based pricing, the customer's going to come and ask to renegotiate. And if you don't do it for them, you're not going to have a happy customer and a decent long-term relationship with them.
So there's something to be said for this dramatic moment of a macroeconomic headwind of these consumption based models losing some of their luster. But we'll go back into another cycle that's more positive.
And ultimately this is just a better value for the customer. It gives them that flexibility, and I think it is more predictable in general. When things are good, it's more predictable. You see exactly what your growth rate is going to be and it holds pretty steady. When things are going badly, you got to admit it's somewhat predictable, it's going to fall in the market. But that's good for your customers and that's what you want long term.
You just take your lumps with that. Who cares what the market says? The whole market is suffering. So it's not like you don't have a big asterisk on your company. You're ultimately flexing with your customer's needs.
KP: This newsletter focuses on growth and go-to-market strategy. What have you learned about scaling go-to-market for a cloud product like Cockroach? What role does PLG play?
Product-led growth has different contours depending on where it's being applied.
There are certain kinds of products where, especially if you're selling to the low end of the market, product-led growth is everything. It's your sales engine and you really want to create a low friction fast time-to-value for that aha moment. And then really feed all of those signals directly into, say, your inside sales team, whatever it is that can process those high velocity deals.
With Cockroach it is a little bit different. We still, believe me, think of product-led growth. In a big customer you might sell a bunch of credits, for example, in the cloud service inside that larger account, and you want to make sure that all of the friction is gone for other teams to try out the product and to be able to use some of those credits. Or if they can't, how do they get a free trial? It's very different to think of a free trial for everyone in the world versus a free trial for a big company that you already have an MSA with that's just going to potentially be able to apply some budget to adding more credits to this big deal.
So PLG is a part of everyone's strategy if you want to be successful.
Thanks for reading Growth Unhinged! Subscribe for free to receive new posts and support my work.
KP: What do you think about the big cloud vendors? Are they partners or competitors? How do you leverage those relationships?
Well, you have to collaborate with them, obviously. Especially if you're building a cloud service.
I found that to differing degrees and in different dimensions, all the cloud service providers have strengths and weaknesses. None is uniquely the best, at least in our experience so far. Some are better at discounting, others are better at co-selling and co-marketing. It helps if they don't have a directly competitive product. But these days what don't they have?
Where you can really win is just finding where you're additive to their overall effort. It's also important to remember that they're very far from monoliths in their thinking. It's easy to have the impression that an organization has consistency in it because that's how a smaller company sees the world.
At these big cloud service providers, there are reps there that are selling 200 services. If they know about your product, they'll be happy to sell it because they're going to get comped on it. (If you have the right co-selling arrangement anyway). But that's the challenge. How do you get someone that has 200 products already to know what the heck your thing is? It's almost an impossible task.
You have to start thinking, okay, what is the specific vertical or the specific geography? You really want to get focused. You want to find a valuable co-selling niche where you can actually do the enablement. At least at a minimum, you need to understand what the problem you can solve that they can't do without you or just tell them what to look for in terms of what a customer is saying that’s actually a better match for Cockroach.
So, how do you enable that? It's a process of finding who's the rep, what's the vertical that we really hit? It's like technology and financial services, let's say. What's the area where we do best in those? Okay, it's going to be West and East Coast. How do we get those sellers that actually are specialized in selling Amazon databases so they already know how to talk databases and just enable that small subset, which is still, by the way, hundreds of sellers.
Part of that is, okay, how good is the partnership with the cloud service provider? And that's a full-time job for somebody to actually try to manage it and make that enablement work.
KP: Do you have any closing advice for readers on your transition to the cloud?
It's worth repeating and it was a big learning for us: you want to throw as much weight as you can behind your cloud efforts, even if it means that you're going to take a short term hit.
What we've found is that cloud services are a tremendously better business model. At least they hold that potential. I mentioned the value to the end customer of the consumption-based pricing. They're paying for exactly what they're using and that feels virtuous and easier to swallow, especially if you start small for companies. You're not absorbing a whole bunch of cost upfront to have some hope of actually building that value by moving use cases and things over time.
For the business, you're attaching yourself to a much larger part of the value. It's like a vertical offering because all of a sudden you're not just selling software licenses in support, you're actually selling the DevOps and all the human labor. That's actually a more significant cost than the license cost, traditionally.
And you're holding out the promise ideally, that as you get scale, you deliver a better total cost of ownership. It’s cheaper for the end customer, plus they're paying for exactly what they're using and they have this ability to scale up, let's say by double. How hard is it to double your cost of software licenses? Not hard, but how hard is it to double the number of DevOps people you have? It's not necessarily one for one there.
Because you're paying for all of these additional costs, whether it's the humans or the cloud costs, you're actually able to think about how do we optimize those at scale, right? Now you're doing this for a hundred customers, not just one customer or every customer doing it on their own. So you can actually get good at that.
Some of it is discounts, for example, that you're driving with the cloud vendors as you get scale. Otherwise it's how do you use your innovative technological capabilities like the thing that made you a leader in your product category to actually make use of the scale you have with innovation?
I'll just give you one example of that that we're engaged in. We traditionally use AWS’s EBS product to store the data. That's quite expensive, but it has really good characteristics for latency and things like that. It turns out that in most database instances or clusters, for most use cases, there's a big chunk of the data that's not very heavily accessed or even at all. It could be 90 to 99% of the data. So it doesn't all have to go into EBS. All of a sudden you can pick a substantial cost and move it into something like S3.
Anywhere there's a big cost in the database, it can be optimized. And so as you have hundreds of customers all under a cloud service, you can find ways to really efficiently use those underlying resources and pass that on to the customer and also improve your margins.
With the cloud service as the customer usage and value grows, you capture more of that value.