How to measure Egineer's Performance
TLDR;
You shouldn't.
Longer version.
How to evaluate the performance of the engineer?
That is a wrong question. The good one would be 'why?'
I was always confused by the question I got from compensation people when there was a new hire: how would you compare existing senior engineer X to this new hire senior engineer Y? Please let us know in % how you evaluate Y to X after the interview.
Fuck I know. And how would you compare even two existing senior engineers? They are adult people with their strengths and weaknesses. So how I will operate with these people and use their skills is essential. Diversity, FFS!
The industry suffers from the crisis; companies do enormous layoffs, and after, they try to make the best of what they can afford. Which by itself is a very understandable and logical idea.
However, bureaucratic and corporate habits do not go away quickly, and I can observe 'history repeating': how industry becomes obsessed with efficiency.
The goal everyone wants to achieve is to improve velocity: release faster. So the method is to ensure everyone works 'well' or 'perfectly.'
The idea on the table 10 years ago was to measure each separately: promote best and eliminate 'bad seeds.' Yes, I am talking about 360 reviews, which were quickly hacked by people who started working for a good review and not necessarily for a good result. Also, the system could be better, as it's very well known that a person struggling in one environment (read team) could prosper in another (literally next-door team).
Recommend reading a short article from DHH: 360 degrees of phony back-patting.
I hope we have passed this period and no one tries to measure individuals outside of feedback from the direct managerial line. Instead, companies try to implement flexible movement between teams, avoiding conflicts between managers for the best people. Healthy move!
Did you notice how easily you understood this: 'best people.' I bet you pictured someone you know from your company or team. It's because, let's be honest, it's transparent: who is good, who is okay, and who 'could be better.' We, team leaders, engineering managers, and peers, know that. It's not rocket science. But we are super lazy, uncomfortable, afraid, or whatever... to do something with it. To do our job.
It's not an article about how to work with underperforming people or how to support overachievers.
I am still conveying the message that a company, as a vast organism, shouldn't measure each tiny part separately and compare them or enforce some methodology for doing things. But, at the same time, we, managers, should care for individuals as a metaphorical representation of the company's 'parts.'
This is a trust issue between the company and the teams. Yes, I know, I write a lot about it. But, if this trust is broken frequently, the company will enforce some bureaucratic procedures because there is no way it can do something on a large scale another way. As an analogy: when you feel your body is out of shape, you go to the gym, and there you do some semi-scientific random shit to your body, hoping it will work. Sometimes it does, sometimes you know.
Because of the chaos of distrust, and lack of transparency, we have all these stupid methodologies we suffer from today.
So finally, we are here: this is the root cause of the desire to measure performance, distrust. When actually, what needs to be done is to build a solid and responsible managers layer, where managers have complete freedom of how they operate, but on the other hand, have transparent processes and a reputation of being predictable and deliverable. But it's long term solution and in crisis it's natural to search for a sliver bullet.
I have yet to see a software company that is happy with the results and, at the same time, implementing productivity measurements or new workflow methodologies. Because why do dramatic changes or introduce new bureaucratic practices if everything is going well?
The intent to measure people's performance shows our managers' failure, broken trust, and failed credibility.