Since reading Radical Candor, I've been thinking about performance reviews and pay scale systems. The punchline is that I don't think that pay and performance should be directly linked, as pay doesn't motivate performance (at least, for knowledge workers, that is). Rather, market rates should determine pay, and performance in turn, will eventually determine your promotion level and hence, market rate. More on this below.
Unrelated, and a quick aside, I still have Voltmeter Clocks on my brain, and plan to write about Vickrey Auctions next week.
Radical Candor's Approach to Level Setting
Radical Candor suggests using a rubric grading system for a twice a year performance evaluation. Some of the features of this suggestion are that the rubric scores can be crowd sourced from an employee's peers and the manager's rating distribution curve can be compared to all other managers. The idea here is to try to use a wisdom of the crowds approach and create a measure of calibration across the company. This way, the problem of a particularly lenient, harsh, or inconsistent manager and their affect on their employee's ratings is reduced, if not eliminated.
In this way, the hope is that performance evaluations are conducted in a fair, consistent, and calibrated fashion across the entire company, and therefore pay raises become fair and consistent as well.
The Problem: Pay Doesn't Motivate Performance
I've written about performance evaluations before, and I'm split on whether I think they're worthwhile, but I do see some value in having a formal process as a forcing function and a coaching tool, and creating some additional material that can be used as a data point for a promotions committee.
The problem I have is with tying performance reviews to pay. For knowledge workers, pay doesn't motivate performance. Knowledge workers are motivated by autonomy, a chance to develop mastery of new skills, and recognition of their peers. In general, knowledge workers don't think much about pay, as long as they're paid fairly and competitively with the job market.
Their are a hundred reasons why performance might vary from quarter to quarter, all potentially factors beyond the employee's control, including challenges of their immediate projects, difficulties with internal and external stakeholders, events in their personal lives, and other factors.
Even with the calibrated review process described above, knowledge worker performance will always remain subjective, difficult to quantitatively measure in a team setting, and is influenced by too many outside factors, and hence is inherently unfair when tied to compensation.
Pay Should Be Tied to the Market Rate
So if performance doesn't determine pay, what should you do? Well, there's one situation where performance is entirely unknown, and that doesn't stop companies from offering employees a dollar figure. That situation is when making a job offer to prospective new hires.
In this situation, the company has no choice but to offer what they judge to be a competitive wage, based on how the company wants to compete in the job market.
If an employee ever decided to leave the company, again, their exact performance would be unknown to their next company. They would be paid market rate yet again.
So why ever create conditions where an employee isn't being paid market rate?
Determining Markets Based on Promotion Level
So if employee pay should be determined by the market, and not by performance, how do we determine what market they're in? Netflix's approach is simply to pay top of market, so whatever competing job offer you get, Netflix will match it, but that can get expensive quickly, and top of market pay isn't possible for everyone (someone has to be top, after all.)
I've seen Joel on Software and others suggest paying people based on promotion level and years of experience. I buy into this, but the next problem becomes determining promotions fairly.
On this point, I think the approach to promotions suggested by Radical Candor makes sense. As with performance reviews, you can't let one bad boss play favorites with their employees and artificially accelerate or stall someone's career chances for promotion. The solution is to set up a separate promotions committee or process outside the control of the direct manager, that similar to performance reviews, calibrates the promotion levels across the company.
A bonus of this method is that managers in turn, can be judged on their ability to grow their employees, and the number of employees they get promoted.
Also, it's worth noting that within this system, high performing employees should be promoted and will be, where their higher market value will equate to higher pay.
Keeping Pay and Performance Separate
My final thoughts on this are about execution. Whenever I've been at a place that said that performance reviews and pay weren't related, there was remarkably little explanation of how pay was actually determined, and the raise cycle happened suspiciously close to, and shortly following, the performance review.
I would follow Radical Candor's recommendation to conduct performance reviews semi-annually. I would conduct promotion reviews quarterly. Promotions would come with pay raises. Finally, I would assess raises based on market rates, either annually, or semi annually, and I would time it to fall before the review cycle, rather than after.
Again, the idea here is to drive towards pay fairness and transparency, and to generate consistency across the company by taking pay decisions out of the hands of individual managers.
To recap some of the big takeaways, I think they are
- Use cross company calibration processes for promotions and performance reviews
- Managers should be responsible for coaching their employees and making them better, not determining their employee's pay or promotion level
- Pay employees a competitive wage, with a fair and transparent pay scheme
I'm still forming thoughts on how to manage top talent and exceptional, star employees, and whether this system works for them as well, but that will have to be a post for another day.
aka THE Awkward Engineer