Daniel Doubrovkine bio photo

Daniel Doubrovkine

aka dB., @awscloud, former CTO @artsy, +@vestris, NYC

Email Twitter LinkedIn Github Strava
Creative Commons License

I recently worked on the Artsy Engineering Compensation Framework, and more specifically on the Engineering ladder, open-sourced today. I have done this before, but never with so much intensity, thought, feedback and so many iterations. This was a very educational exercise for me and I wanted to share the process surrounding the giving birth to something like this.

I began by asking the New York CTO club.

Now that my team has reached a certain size I am starting to feel the need to developing a compensation framework. I am not even sure what that should look like.

I got a wild variety of responses. It was very clear that there was a lot of merit with having a solid base in terms of positions, levels and job descriptions. Those gave people a clear understanding on how they fit and what their future goals might have been. To actually do the math bigger companies seemed to create formulas for calculating salaries and a few names were dropped, including Tower Watson or Compa Ratio. All these systems seemed to inadequately reflect people’s expectations, particularly felt in regional differences. In the end I concluded what I had already known: compensation is more art than science.

I started looking for other technologists going through the same exercise at similar-sized companies. Camille from Rent-the-Runway shared her ladder with me (and has later open-sourced it). I edited that and created two documents: an Engineering Ladder for individual contributors and an Engineering Leadership Ladder for team leads. I had those diverge at mid-point: I typically ask a strong individual contributor whether they would rather focus on developing people or whether they would like to focus purely on technology, then help them grow into either direction. I also made a point to emphasize that a team leader at Artsy is also a very strong individual contributor, leading by example. Finally, I tried to use the ladder as a checklist, add actual numbers to yield compensation ranges.

I presented my two ladders to the rest of senior management and got a lot of negative pushback. My ladders ignored company values and potentially could have yielded a very adversarial environment for negotiating pay. I tried to work the values part into the ladder, but that made no sense - did an Engineer 4 really contribute more to company values than an Engineer 3? Having spent a lot of time on this I was discouraged and beat.

One of the suggestions made during an early review suddenly clicked for me.

Make performance evaluation a combination of the what, i.e. what you achieve that contributes to your team’s goals and ultimately Artsy goals, and the how, i.e. how you act and how those actions contribute to Artsy’s culture and values.

The what part for an engineer would then be expressed in an individual contributor ladder, which could describe typical growth from someone who has recently graduated to an inspiring technologist. It could also be replicated to other teams and list the specific elements of an individual contributor’s job description. The how part would be entirely left to your team lead, and would be defined by the company as a whole. The leadership ladder was no longer necessary, because that could be fully grounded in literature, specifically in Tribal Leadership for Artsy.

Armed with 5 individual contributor ladder levels and numerous compensation surveys, anecdotal market data and clever use of recruiter spam I was able to come up with a fair market compensation range for each. I liked how that represented typical individual contributor growth and typical pay. I also made it clear that this is a benchmark or a reference, not a formula and that value-based compensation decisions can get someone closer or even above their high market value. If you run an Engineering team and would like to share numbers, please e-mail me.

Determining actual compensation for each individual posed some challenges. The first one was that Artsy often offered a range of compensation to every engineer. For example, the company may have offered you a choice between a) $80,000 with 15,000 stock options, b) $100,000 with 10,000 stock options, or c) $120,000 with 5,000 stock options (these aren’t real numbers). If you could afford it, you were buying an option to invest more into the company in the future with your salary, or you could have chosen to invest less for more cash upfront. Instead of thinking about total compensation I followed a suggestion of subtracting or adding the cash difference to the target market comp. In my example your ladder was, say, Engineer 3 and the fair market value for Engineer 3 was b), hence the offer above. Had you chosen the c) option your fair market compensation became $120,000, a difference of $20,000, acknowledging that you wanted more cash and were willing to forego some stock options for that at the time of signing. We would also continue giving you subsequent stock grants over the years with your choice in mind, always aiming for all employees to have a fair amount of stock based on a similar framework, in order not to break the compensation model and generally retain people.

We tested this framework against the team’s current compensation, salary increases, and in 3 offers that we made between January and March. An interesting example came up with a candidate whose compensation expectations were below our market value - the framework told us to make a higher offer making one very happy new employee. We also had the reverse and made a lower offer, but were able to justify it to the candidate and feel that we were consistent and fair to every single team member with our offer. We had no trouble closing all 3 exceptional engineers.

This was a real and a very interesting challenge. If you haven’t read it yet, find the Engineering Ladder at the bottom of my blog post on the Artsy Engineering Blog.