Why your organization’s promotion process is broken and how to fix it.
Why are your best employees leaving? Read on for my overview of the different methods that promotions are driven and how they cause dissonance.
Method 1: Manager driven
This is a parliamentary debate in which the managers of different units sit in a room advocating for their team members and who should be promoted. A board of cross-functional technologists/ managers take the call on whether the work done is really worthy of a promotion.
What’s broken?
Don’t even get me started here. There are a whole bunch of problems.
- Your promotion depends on your manager and his selling skills.
- Some managers are more aggressive than others and are able to battle it out better.
- Your promotion depends on your relationship with your manager. Unless there are a significant number of levels separating you and your manager, he might get threatened.
- Your manager might try to maintain egalitarian status in his team across all his developers to make sure that employees don’t grow too fast. The crappy “Myth of the 10x developer”[4] has caused people to believe that it is okay to treat good developers on an equal footing with everyone else.
- Your manager might push experienced developers instead of less experienced ones.
- If your manager has a different skillset than yours it would be difficult for him to understand your capabilities. He might promote people whose capabilities he understands.
- If your manager’s appraisal is determined by the number of employees he is able to retain, than he would promote an employee who is inclined to leave.
- Too much power in the hands of the managers.
What’s good?
With a capable manager, this gives good results and helps retain the team which is the priority of the organization. This almost never happens
Method 2: Peer/360 Review
In this case, you try to get a peer review done for all your developers. Everyone gets a say on who will be the the person to be promoted. Democracy, Yeah!!
Unfortunately this is probably the worst way to organize a promotion process.
What’s broken?
- Different teams might have different sizes. This severely disadvantages people of smaller teams or gives them a huge advantage if the feedback is inversely weighted to the team size.
- If the teams are working in isolation, the above problem is severely exacerbated.
- It also gives a disadvantage to people who work in high performance teams where people are threatened by each other’s performance. Attaining peer respect in talented teams is roughly next to impossible.
- Game theory would suggest that peer reviews would result in sub-optimal feedback for the people involved. The below example (a prisoner’s dilemma) suggests Brown and Anderson, two peers would go negative on their feedback for each other if they are competing. (In the real world their feedback would be neutral with left handed compliments).[3]
This example assumes that promotion is a zero sum game among peers and hence this would never work in the real world. However if the promotion across levels is not a zero sum game then this could still be useful for getting manager evaluations as a skip level feedback.
Also your best effort to anonymize reviews might not work out and lead to a conflict within the team.
Note Peers often will not have the necessary experience to rate their colleagues on their performance.
What’s good?
This can be a good auxiliary feedback to my preferred mode which I have mentioned below.
An overwhelming number of positive/negative reviews can be used to uncover insights which would otherwise be hidden.
Method 3: Individual Driven (My preferred option)
The third option which I recommend is an individual driven mode of promotion where each individual in question prepares his own dossier on why he should be promoted. The dossiers themselves are anonymized so that the board cannot determine whose dossier it is, effectively resulting in a double blind mode of operation. Although the manager has a half point say in it, the entire dossier is run through a skip level evaluation with a cross functional board which decides who is eligible.
What’s good?
- This to a great extent removes the managerial bias in promotions. Your manager’s insecurity is not a factor in determining your fate.
- The aggressive and sell-worthiness of managers are also eliminated here.
- This is completely merit driven.
- Each level could have its own set of candidates and compared to other projects which are similar.
What’s bad?
Written communication skills play a big role in your dossier preparation. If you are a prolific blogger like Addy Osmani[6] or Illya Grigorik[7] for example , your promotion would be fast tracked.
There could still be misses. Google follows this process, sometimes with mixed results.[1]
An easy way to spoil this process is by making it purely metric driven instead of having the promotion board spend time going through each and every dossier in depth. This was the gist of how google messed it up above.
What’s missing?
I believe there needs to be a gender ratio to also take into account here to make sure the opposite sex is not disadvantaged. The Software industry has a skewed gender ratio.[5]
Also, In method 3, there can be a way to allow the individuals to get an appeal for their promotion if they don’t agree with the feedback on their dossier. An appeals court if the shoe fits.
Also in method 3, there could be a regional ratio in addition to a gender ratio that can be maintained for promotions across regions. I cannot stress how critical this problem is for International companies where there is a risk of everyone outside the core hq being treated not on par.
Don’t follow Neutron Jack’s bell curve grading blindly. It just does not make sense when you have a high performing team.[2]
I would caution moving people who would be interested in an individual staff engineer track over to a managerial track. Most engineers unfortunately do not know this and do not appreciate you as a manager making this decision for them.
Some companies especially those which have software arms as part of a larger industrial organization try to adopt a pure metric driven approach; for example, code coverage, number of bugs, number of unit tests and so on. Sorry to burst your bubble, but it requires a great deal of bean counting to come up with these metrics and would be impossible unless you are hiring full time project managers ( a designation and a practice that has fallen out of favor because of lack of value addition). Also estimations are almost never accurate.
So how will you run your promotion process now?
