Jay Kreps is co-founder and CEO of Confluent. Prior to Confluent he was the lead architect for data and infrastructure at LinkedIn and the initial developer of several open-source projects, including Apache Kafka. He spent one year in high school before deciding to teach himself—and convincing a dozen of his friends to follow suit—but eventually opted for college.
Empathy. I try to put myself in other people’s shoes and really get a sense of their experience. I used to approach talking to someone as just saying what’s on my mind, but eventually I realized I should be focused on what’s on their mind. This is really hard because in a sense our lives are a bit like a movie we watch, and we see ourselves starring in every scene. The hard thing is to really understand that everyone is watching a similar movie on the same subject but with a different main character—themselves. Companies do big things together as a team, and to get this to happen you really need to understand how everyone sees things.
I also try to assume good intentions. If you assume bad intentions, even if you don’t say so, that will always come through. And maybe a person is truly terrible, but it’s far more likely they’re just trying to solve for something you’re not. This happened to me often early in my career; I would be aggressively pushing for a project and I’d start to see the people who were standing in the way as opposed to progress. But of course, they weren’t anti-progress. They just had different priorities.
The very beginning was the hardest part for me, because I had to adjust to making decisions with so much uncertainty. As an engineer, there’s usually a right answer. If you don’t know what it is, you just need to think about it more or go gather more data. So truly not knowing what to do was really stressful—especially because a young company is such a fragile creature. And you’re making all these really critical decisions and you’re imagining all these potential competitors and competitive threats and trying to figure out how you are going to try to set yourself up. You start to worry about all these things you can’t control.
One of the things that’s been helpful to me, in terms of making good decisions but also emotionally, is to focus on playing the hand I’m dealt. This is a common expression, and is usually used in kind of a fatalistic way, but I think there’s a lot of wisdom in it. If you need to make good decisions with imperfect information in a domain with a lot of luck, you need to change your focus from “will I win the hand,” which you can’t control, to “did I play the hand well”, which you can. I can’t guarantee good results, but I can capitalize on the advantages I have and at least know I’ve done my best. In one respect this is kind of obvious, but in another it is a very different way of judging your own performance.
I never thought I’d like having a routine, but I really do. Early on in my career, I had a lot of time and used that as a substitute for planning—but it wasn’t a very good substitute. Then when I had kids, I was suddenly extremely time-constrained. I had to wake up when they did, and that meant I had to go to bed at a reasonable time. I started to exercise and eat at certain times. And I found the pattern and rhythm made me more productive and much happier.
The most important thing once you realize you have a fixed time budget is prioritizing important things and cutting less important things so that you can produce more. I was managing a team and working on several different software projects, but the most important one was Kafka. So I started putting almost all of my time into that. I actually think most “personal productivity” wisdom like “inbox zero” get it very wrong. They focus on cramming more things into the day, but at some point, it just doesn’t fit and all your work suffers. I think the real way people with high output produce more is not by cramming more and more tasks into each day but by focusing their time on the highest impact things.
More about China. Growing up, we studied European history, and I didn’t learn much at all about Asia. But China is really important if you’re running a company—it’s a major world economy, and our open-source code is actually used twice as much there as it is in the U.S. I’d like to have more of a background in the history and culture, and to better understand what it means to do business there.
I’m just starting to educate myself. Maybe because I come from a software engineering background, my instinct whenever I want to learn something is to find a book, but that’s not always the best approach for every subject. I remember once I was reading a book on exercise and my wife was like, “You know, maybe you should just go exercise.” But I wanted to know the theory of it first!
One I finished recently was The Book of Why, by Judea Pearl and Dana Mackenzie, which is about integrating the idea of cause and effect into traditional statistics. So much is built on the foundation of data analysis, but we usually think only in terms of correlation, not causation.
I also just finished Battle Cry of Freedom, by James M. McPherson, which is about the Civil War. It was fascinating—such a pivotal point in U.S. history, with the horrible shame of slavery but also the war that ended it. And it’s an amazing leadership story, as well. When Lincoln became president, he’d served one term in the House and had no military experience. But that was actually the smaller jump, compared to going from a few months of formal education to being a lawyer and a lawmaker in the first place. He was super smart but also curious—he had what today we’d call a growth mindset. He became a much stronger military strategist over the course of the war, which required military strategy and organization. But since both sides were democracies, everyone had to keep the population in support of the war in the face of enormous casualties. Lincoln managed to transform the war from being about “union” to being about freeing the slaves.
I think something we were wrong about very early on at Confluent was the role of services—helping customers be successful with the product. The common wisdom at the time, which I still hear from some founders today, is that services weren’t scalable and would take away from revenue. People told us to minimize them as much as possible, and that’s what we did.
But after a couple of years, we realized that was totally wrong. Services revenue doesn’t take away from product revenue; if anything, it adds to it. Most thinking in sales is oriented around the first deal you make with a given customer, but in a truly healthy business, at least 90% of the value comes from expansion and renewal. So the organization should be focused on a successful customer journey. Services are still a relatively small part of our business, but that mindset shift has transformed the way we work.
In terms of business, I’d say the long term. Humans tend to measure progress day to day. But at that scale, moving toward something really valuable doesn’t feel that much different than moving toward something with far less value.
What “long term” means will depend on your situation. We started working on Kafka 10 years ago—5 years before we founded Confluent. That might have been overboard, but it did give us a lot of time to think about what we wanted to do and why it was valuable. Even in a situation when your horizon is much shorter, I think it’s still important to think as far ahead as you can and solve backwards for what you need to do now.
The direction will always evolve over time, too, as you learn more about the problem and what’s required to solve it. Whatever you build first will probably be very far from where you end up. So once you get going, it’s also important to carve out time for things like offsites and even just unbooked time on your calendar, to make sure you’re not letting the urgent trump the important.
- Assume good intentions
- Accept uncertainty
- Think (very) long-term