As readers of this blog already know, I’m constantly on the lookout for innovative compensation approaches. “How to redistribute the value generated by the organization to the people who created it?” is one of the most profound organizational questions. And financial compensation is one of the most tangible indicators of our values and beliefs systems. Any attempt to shift to a new operating paradigm without taking these two issues into account is bound to fail.
Over the last several years, Nicolas di Tada and the team at Manas Tech, a 30-person Buenos Aires-based dev shop, have carefully evolved their process for allocating pay raises. Not only did they document and share their process in:
But Nicolas was also kind enough to hop on a call with me a few weeks back and clarify some of the points that were not clear to me at first read.
In a traditional compensation review process, an autocratic decision-maker (manager), uses quantitative inputs from a performance review to set a new salary according to a predefined salary ladder. The team at Manas sees challenges, bias and limitations in all three key “design elements” mentioned above, so they set out to design a compensation review process without them. The key design principle underlying their system posits that the “wisdom of the team” would lead to a superior outcome than a process using the three elements outlined above.
Their process currently works as follows:
- Every 4 months, the team will review its automated financial model to determine the portion of profit that should be allocated as salary increases. If confidence in future billable hours is lower than desired, the same amount will be allocated as one-time bonuses rather than permanent salary increases.
- The process runs in 3 to 5 rounds (exact number determined at the beginning of the cycle).
- In each round, each team member sees the base salaries of all other team members, and the total pool of salary increases that can be allocated. They can then allocate it across team members in any way they see fit. Team members cannot give themselves a raise.
- At the end of each round, the average increase that each team member received gets permanently allocated to them and subtracted from the overall pool.
- The next round follows the same steps with team members also being able to see the cumulative salary increases that were already permanently allocated to each team member, and the updated (lower) total salary increase pool still remaining to incrementally allocate.
An interesting challenge from a technical/algorithmic perspective has been dealing with the “fuzzy” relationship between an individual team member’s input/recommendation for a salary increase and the resulting increase, since the recommendations of all other team members have to be factored in as well. It creates an incentive to provide an input that’s different than the outcome that you think is appropriate, in an attempt to account for the impact of the other inputs.
In order to get as close as possible to the desired outcome, the solution that the Manas team landed on after multiple iterations is to only permanently allocate the average recommended increase and run several rounds of the process.
While the process does offer a unique solution to some of the greatest challenges of the more conventional approaches, it does pose its own set of challenges. The two that immediately come to mind are scalability and market dynamics.
The solution works well now for Manas at ~30 people where most people know most people well enough. But what happens where there are 200 people in the org? Simply averaging the increase recommendations in each round will require a lot more rounds since a smaller portion of teammates will have a non-0 recommendation for each individual teammate. Potential solutions can be either finding a more permissive “discounting function” that’ll require fewer rounds, or potentially following a tiered process where execs allocate the overall pool across departments, managers allocate the departmental pool across teams, and individuals allocate the team pools across individuals. Each of these comes with its own set of advantages and disadvantages.
The market dynamics tension is a bit more challenging to resolve and the Manas team hasn’t found a one-time systemic solution to it. If we evaluate the Manas approach through a “compensation polarity” lens, their approach falls very close to the “internal fairness” pole. Since compensation market data and market seniority definitions (levels) don’t play a part in the process, it’s not unlikely for salaries to drift overtime from their market comparables and have someone in a role where they are being paid either significantly above or below what they would get paid for doing a similar role in a different company.
In sum, I’m grateful for Nicolas and the folks at Manas for taking a pretty big leap, redesigning a compensation system from scratch breaking many of the challenging assumptions in a more conventional system. It is not perfect and not without its shortcomings, but neither is the existing system, making it a viable alternative offering a plausible trade-off.