Designing a bonus plan
I once had to define an employee bonus plan for the company I was working for. I struggled and the outcome was less than optimal. Here is what I learned.
TL, DR: I came to the conclusion that a company shouldn't have a bonus plan.
As put very well in "It doesn't have to be crazy at work", if you have the possibility of a bonus, people come to expect it and are disappointed if they don't get it. In other words, the best outcome you can hope for is that someone is not disappointed. It's much better to pay people well and avoid the stress that comes from the anxiety of not getting the bonus.
Why do companies want to have a bonus plan in the first place? The conventional answer is "to motivate people to work hard and do what is important for the company". My experience is that the effect can be the exact opposite. Smart and passionate people know how to motivate themselves and do what's right, but might actually give up their judgment if they know that their compensation is tied to some, likely already obsolete, targets.
I made the mistake of tying bonus payments to "Objectives and Key Results", and it was a disaster. People pushed back on challenging objectives, plus the real goals for the company changed almost immediately, making the targets obsolete. At the time of the bonus target review, I had to make a lot of excuses and explanations to reward the people who had had the biggest impact, even though that impact wasn't aligned with the original targets. We tried to fix this by reducing the target setting period to six months but even that didn't help, and caused a lot of extra overhead. I learned the hard way why every book about OKRs recommends that they should not be tied to compensation.
Defining the actual targets is also challenging. When working with innovation and new product development, it is nearly impossible and counterproductive to define measurable targets. It's another thing if your job entails doing something that is clearly measurable, like cleaning X square meters of a building's floor, or process a certain number of applications, or even achieving a certain amount of sales. But when your goal is to launch a new product that you have only vaguely defined, how do you measure the success? How do you hit 30, 50 or 75% of this goal? If the product is 1 month late but is a great success and brings a lot of money, have you met or missed your goal? By how much? And so forth. Not having a bonus plan removes all of these questions and more, and lets you focus on the work.
If you do want to reward people, do it by performance and not meeting certain specific goals. How do you measure performance? Well, the ugly truth is that you just know who is contributing more than the others and should be rewarded. But beware: I read in Masters of Doom that at id Software, after John Romero had left, the owners decided what bonus to give to each employee based on, basically, their preference. This created a highly competitive and unhealthy environment where everyone was just trying to outperform and look better than the others.
But what are goals for then? The purpose of goals is to keep people aligned, not to measure their performance. Everyone's tasks should be in support of a goal, and if they aren't, they should stop them. A goal is a tool to let you question whether something is really needed or not.
At Nokia, bonuses were awarded based on both company and personal performance, and there might have been an organizational level component but I don't recall it. The personal part was based on a grade given by the manager. While we tried to have targets, they were loosely defined and in the end the manager's judgment decided the score. The bonus percentage was anyway negligible, and the personal portion (I think 30%) even more so.
At Intel the bonus was based on a complicated formula that took into account financial results, company goals and organizational goals, but no personal component. Nobody was able to predict what the bonus would be and how to affect it, so everyone just went along with their work and hoped for the best. Intel's bonuses were more significant than Nokia's, but also salaries were higher in general, so getting the bonus was a nice surprise but not getting it wasn't a big deal.
Contrary to popular opinion, I think that if you must have a bonus plan, then tying it to the company's results makes sense. It may feel like a lottery and people might not see the direct connection between their work and the outcome (although if they don't, you have another set of problems), but I at least can relate to the idea that if the company is doing well we all share in the results, and if it isn't we tighten the belt a bit.
But no bonus is still the best option!