James Cowling

7 days ago

Company Values that Matter

What makes a good set of values? Do companies even need them? Aren’t they just empty platitudes that achieve little more than eye-rolls from the engineering team? Yes sometimes, but I think a good set of values can be critically important in establishing a team that stands for something.

Values underlie everything we do

I’ve spent a lot of my career getting pulled in to resolve conflicts in engineering organizations. These conflicts usually involve two parties who are smart and well-intentioned yet think the other party is neither of these things. I rarely meet engineers who are dumb and have bad intentions. There must be something missing.

My model for human decision-making looks something like the following, where each layer depends existentially on the layer above:

Engineers spend so much of their job dealing in the “how” that they sometimes forget they have a set of values deep in their psyche that critically influence everything they do.

If one group believes in shipping fast and reacting to user feedback while another believes in correctness and a simple feature set then they don’t have values alignment. They can argue all day about whether to add another special-case API and never see eye to eye.

It doesn’t matter how hard people try to do the right thing if they don’t agree on what’s right.

So what makes a good set of values?

Good values should involve a tradeoff

If the opposite of a value is nonsensical, then the value is meaningless.

I loved it when Facebook had the value “move fast and break things.” I didn’t interview there because that’s not what I value, but it’s a legitimate tradeoff I can respect. Then years later it gained a few syllables with “move fast with stable infrastructure” but lost a whole lot of meaning along the way. Do you want me to move fast or slow down and get it right? I don’t know many companies with the value “move slow with unreliable infrastructure”. It might as well just say “be better at stuff.”

A value needs to involve a tradeoff. It’s meant to guide decision making and needs to capture an actual opinion, an opinion you’re willing to stand behind when times are tough.

Each company value at Convex comes with a set of “intentional tradeoffs” illustrating some of the costs we actively accept. Take a look at the tradeoffs in our “Make Work Human” value. It sounds positive on the surface, but it also means people need to come to the office in-person, need to put up with distractions, sometimes need to keep their controversial opinions to themselves, and shouldn’t look to me to tell them how to think about a political issue or global tragedy. These are certainly not universal values at modern corporations.

Good values should not be aspirational

Violating your values is a big deal. You should probably feel bad. So they obviously need to reflect the current reality of your organization. If you’re not even acting in accordance with your values on day one, then you likely never will.

Don’t try to use values to “reboot” your company culture. Get your house in order then codify the status quo in a set of principles that reflects who you are, not who you want to be.

The value “Do it for the why” is deeply ingrained in our DNA. We talk about it every planning meeting. Our product is very intentionally designed and everyone here knows you can’t get away with working on a project without a crisp articulation of why you’re doing it and what you hope to achieve. “My manager told me to” is not a satisfactory answer… even if I’m the manager.

If you’re a data-driven startup thrashing around and iterating your way to success then that doesn’t fit our definition of “doing it for the why.” If you’re a command-and-control style organization where the team just does what they’re told, this is also not how we operate. We all have different styles, you do you, but pick values that uniquely capture who you are and how you operate.

Good values should not be designed by committee

Remember the first paragraph when I said values are about standing for something? It was only a couple of minutes ago. Unfortunately this means you can’t just send out a survey and crowd-source your values.

Values underlie all decision-making at your organization. They’re the first step towards rallying a team around the “why”, “what”, and “how”. You have to actively drive this process and own the outcome of it. The values also need a cohesive voice and to fit together into a set of guiding principles, rather than just a grab bag of opinions.

Everyone at Convex gets behind the “give a shit” value but they may not have suggested it. It’s not easy for someone to suggest that the best way for them to grow is to prioritize the mission over themselves, but it's what we believe here. Sure the profanity in the name is a little tacky but we didn't want to pull our punches with our commitment to the value. It’s hard to reject job candidates who are world class in their field but don’t actually care about our product, yet we do this all the time.

We stand behind these values and I wouldn’t expect the team to respect them otherwise.

Good values should be relevant to the business

This isn’t just another virtue signaling exercise, the values need to be directionally relevant to the business. Being true to them should help your company succeed and ignoring them should hinder your success. Success means more than just making money of course, presumably you also want to create something of value.

The impact of company values usually only shows up on long timescales. If done right they’ll have a lasting cultural impact and guide your company long after you’re gone. If done wrong it’ll take a long time to notice and then they’ll be very hard to change. Values have inertia.

Convex places a high importance on being intentional and passionate about our work (do it for the why, give a shit) but as a result it can be easy to spend too much time debating instead of actually building. This is why we have the “drive forward progress” value. It recognizes that we’re a business that needs to ship, that talk is cheap and doing is hard, and that the company will reward those who are brave enough to push us forward.

Good values should be meaningful

This is the most important part; good values have to be significant. They need to resonate with the team. People should get upset if they see them being violated.

The most important Convex value is “Be Worthy of Trust”. We stole that one from Dropbox... but like… in a trustful way.

It’s easy to be cynical of big companies but we really lived that value at Dropbox. Being worthy of trust was hard. We spent huge amounts of money and effort doing things like building a storage system that never lost a file ever. I can’t tell you how many weekends we worked and social events we had to skip. I also don’t think I ever heard someone complain about it. We were doing something we cared about, and the fact that everyone else cared about it too made it feel like a movement. That was a great job.

Young Sujay interrupting a bike ride to respond to a production issue, a decade before cofounding Convex. I am helpfully taking photos instead of helpfully helping.

Picking good values

Picking values is meant to be hard. It’ll take time reflecting on what your company stands for and what’s unique about it.

The Convex co-founders have very different working styles but are strongly aligned on values. That alignment is why we work together, and is also the only reason Jamie and I haven't tried to strangle each other. When we started the company we even wrote a “founder marriage” doc where we restated our personal values going into a startup to make sure we were on the same page, despite having worked together for eight years.

We didn’t pick company values until a few years into running Convex and all our youthful optimism was stripped away. The rough process was:

1. The founders met on a series of weekends when there wouldn’t be any work distractions.

2. We individually brainstormed values on post-it notes then tried to group them into clusters on the whiteboard.

3. We got lunch. I caught covid. I ate pasta, it was good.

4. We wrote down a set of cultural risks that could lead to the business failing. This part was easier because we’re engineers and engineers are good at being negative.

5. We cross-referenced the values and risks and asked if the values were necessary and sufficient to help the business succeed in the way we want it to. Then we tried removing and merging some of the clusters and repeating the process.

6. We stared into space for a long time then repeated step 5.

7. See step 6.

8. We ended up with five clusters and came up with a short phrase to capture each one. This was before ChatGPT so we had to use our brains to do this. You can ask your grandparents what that was like.

9. Over the next few weeks we interviewed engineers at the company what they think the values should be. We didn’t want to pick values by committee but also wanted to make sure there wasn’t any misalignment with the team. Turns out we were aligned. If you’re not you might want to bail out and fix that first.

10. We each took a shot at writing descriptions for the values and a list of downsides we’re willing to own in order to uphold them. We had a single person write the final descriptions and tradeoffs so they’d have a consistent voice. Then Sujay went and changed all my punctuation because he’s pedantic like that.

11. We had a company meeting where we talked through the values, answered questions, discussed tradeoffs etc. It went well. It would have gone less well if we didn’t take our time getting to this point.

12. I forgot to write a blog post about it for two years.

13. Well here we are.

This process was as arduous as it sounds. Unlike many other management activities however we all came out of it glad we spent time being systematic about what we value, and proud that we have a set of values that the team actually cares about.

Hopefully your values process is similarly useful to you, and hopefully you also end up with something opinionated. Ideally your values will help attract employees you want to work at your company and ideally it’ll turn some people off too. Both of those are great outcomes and will aid in building an organization that, above all things, stands for something.

A huge thanks to Mindy Zhang who got us pointed in the right direction with some great advice when we started defining our values at Convex. Thanks also to the early Dropbox crew who created one of the strongest engineering cultures in the industry and influenced us in many ways.