Do you want to build a Snowflake?
Notes from Frank Slootman's talk @ FC Build
Apologies in advance that I haven't been as regular lately. I expect to be on and off for at least a few more weeks and then get back to being regular again. Little humans keep us really busy :)
As some of the readers here may have read, I've been following the Snowflake journey quite a bit on this substack. Yesterday, I had the privilege of hearing Frank Slootman in a fireside chat at the FC Build event and I took copious notes that I wanted to share with the broader community. Thanks first of all, to the good folks at Foundation Capital for hosting this and inviting yours truly. Each word by Frank was pure gold - he is definitely a straight shooter. The discussion covered a gamut of topics and I'm going to break them into sections (quite arbitrarily). A lot of it was also covered in Frank's essay Amp it Up.
Success and Failure
Be paranoid, not celebratory - we are all playing in markets where its dog eats dog and incumbents will fight hard. We have to always stay paranoid and make that paranoia work for us - keep us on our toes
Don't celebrate too hard. Frank mentioned to the extent that he probably celebrated only 5 mins for the IPO
We all live in a world where anxiety is very common. However, just like paranoia, we have to use that anxiety to bring focus and sharpness in our actions. Embrace the anxiety and use it to fuel your concentration. Get comfortable with being uncomfortable
Hiring and Firing
Hire for aptitude. Experience can be learned but aptitude is hard to teach. Engineering hiring often falls in this bucket, but harder to replicate in other areas. Up and comers.
Always keep asking the peer group to evaluate others and if the peer group is not excited about somebody who is not pulling their weight, its time to let them go. Over time he started to move faster and faster as he got more experience and built confidence but he has never felt that he moved too slowly in a matter of firing.
Personal attributes like race and gender never play a role in evaluation. It's all based on aptitude and data. He made a joke that the only thing he doesn't want is more people of Dutch origin :)
Leadership
Leaders have to set the pace and tempo. An organization by itself will move at a glacial pace (he gave the example of DMV). Leaders have to raise the bar on outcomes by challenging everybody to get things done faster. If somebody on his team says that they will get "X" done in 6 months, he often counters by an order of magnitude lower - why not in 6 weeks, or why not tomorrow afternoon? In general, people will elongate timelines and leaders have to set the pace. Speed is energizing and electrifying.
If you are presenting something to a leader, you need to be thrilled beyond bets about it. High energy and conviction is the most critical part of getting leadership alignment. And Leaders are the energizer bunny of the whole organization
You have to drive customers too since you need to make them consume - that's the only way to recognize revenue esp for Snowflake.
He wants people to push him instead of having to push them. They need to be very confident and secure and not afraid of him first and then push him. Don't come and ask him what you should be doing, instead tell him what he should be doing.
Leaders need to have a very high clarity of mind and keep reinforcing it at every turn. What are they doing it and why are they doing it? Otherwise, by the time it goes down a couple of layers, there's mass confusion, almost like people freelancing and running their own show. It's a bit like 5 years old playing soccer - a bubble moving across the screen and nobody holding position.
Build very high conviction. If there's no conviction, there's a problem. Can't be wishy-washy. For instance, while hiring people, you need to move forward with a lot of conviction else it doesn't work out.
Intellectual Honesty. See things how they really are vs what is politically desirable esp when things are not going well. In a lot of cases, people try to step around the main issues and try to discuss other things with their leaders instead of addressing the core problem. Frank gave the example of how doctors spend almost all of their time in diagnosing and very little time in therapy - if the right problem is identified, the therapy is easy. Often, leaders will jump to conclusions and have their favorite narratives, and spend a lot of time on the "therapy". For instance, fire your VP of sales when there's a product problem. That's intellectual dishonesty and happens in VC backed companies all the time.
Org Design
There should be a singular metric that has to align the whole company. People have too many priorities. He likes to talk in singular and not plural. He hates MBOs and Balanced Scorecards - gets rid of them everywhere it goes. Stay hyper-focused on one metric. At ServiceNow, it was net new ACV (accounting for attrition, expansion, and new accounts) and at Snowflake, it’s all about consumption since they only make money on consumption. He gave an example of sales getting paid commissions on bookings but the company ran on consumption. The whole organization has to align on one single metric.
Make engineering aligned with support. Make sure that engineering is sensitized to the pain customers face. Engineering can "solve" the pain but support can only manage it.
While they have NPS scores, they are hyper-focused on the narrative behind the NPS scores. Very aggressive in chasing out any narrative where customers are not happy
Founder → CEO Transition
(I felt a lot of this also applied to leadership in general)
Many young CEOs hire senior talent and then "delegate" — sit back and wait for things to happen. E.g., somebody would hire a VP of sales and then wait for the business. Which leads to a disaster. You need to have an "active" relationship. Weekly check-ins in which you are also interrogating, should be an active exchange and collaboration/co-operation. It's the CEO's job to make sure this executive is onboarded. The word Frank used was "driving" not "presiding" which is such a great way to describe this relationship
Founders need to understand their relationship with their board and the boundaries that exist between the board and CEO. Assert the full scope and measure of their role vs surrendering to the board. He gave an example of a founder telling him that the compensation committee is imposing their compensation philosophy on him.
In most VC-backed companies there are smart people in the room. If you leave a vacuum, they will fill it up. Don't say "What do you think?" Always go with a clear direction of what you think and go with conviction. He wished he could give this advice to his younger self.
Covid & Cloud Trends
Covid became the catalyst for data initiatives going up in priority in most enterprises so that has been a big positive. Snowflake has zero on-prem equipment so they just continued operating exactly the same as pre-pandemic. In the Mar-Apr, they deprioritized some roles which were far removed from the "drive train" (core engine that generates revenue) - and lightly touched brakes in a few areas
No excuses - pretend with the team like it doesn't exist. Once you start worrying, "Oh, there's a crisis", you start to condone non-performance
He remains excited about Data warehousing and the fact that data is fragmented into a million pieces. Snowflake shines in terms of scale, computational performance, and concurrent execution. Alongside, it’s building a federated architecture to create an orbit of data making it easy to share data within and outside the organization (I missed noting a few things here).. ultimately sees it as a "data cloud"
He talked about the concept of the Time Value of data. Data loses value very quickly and latency is super important. Data is valuable in seconds and minutes; hours and days are too late. Google and Facebook pioneered the idea of "Data Driving vs Data Informing" - snowflake democratized this for the rest of the world (Aside: Snowflake market cap today is greater than the TAM ascribed in the S-1)
He believes that there's a tremendous opportunity in the value-added processing of data. Gave examples of ML models that bring value out of clinical data, image processing, detecting emotional content from support tickets. Every software company is a data company.
Final Tips
Obsess over your product till you can't obsess anymore.
Fight distraction every step of the way. Say no a million times.
Need a 10X+ better product so that the average salesperson can sell it and hit it out of the park. If only the founder can sell it, it’s not likely 10X better.