I’ve held the opinion for many years now that what you might call the “old-school” or “checklist” annual performance reviews are ridiculously ineffective. One might even use the term “asinine”, and I do. There are two main reasons for this: you might call these the “Comparison Problem” and the “Absolute Value Problem”.
The Comparison Problem
I remember the old MSFT performance reviews from the mid-90s: rankings on a one-to-five scale, with one being a cold corpse and five being a superman. Ones and fives were so rare as to be almost unused, and you didn’t hire a lot of people who were natural twos, so 95% of the people got threes or fours.
You’ve heard of Garrison Keillor’s Lake Wobegon, where
all the women are strong, the men are good looking, and the children are above average.
Have you heard of the Lake Wobegon Effect? This is
the human tendency to overestimate one’s achievements and capabilities in relation to others
i.e. we all think we’re better performers than average. So let’s take the Microsoft example quoted above. Let’s say that the “average” score for the company, given that there are only two practical values on the one-to-five scale, should be 3.5. Here’s a hypothetical approximation of the results:
- 25% of the employees (the A performers) get a 4.0, higher than average
- 50% of the employees (the B performers) get a 3.5, the average score
- 25% of the employees (the C performers) get a 3.0, less than average
But due to the Lake Wobegon effect, every employee walks away from the performance review dissatisfied.
Even with subconscious grade inflation by the managers (which anecdotally happens A LOT) you’ll still have a lot of dissatisfied employees. It is REALLY REALLY HARD to pleasantly surprise a person come review time.
The bottom line is that the Bell Curve works against you when it comes to performance reviews and employee morale. The reality is that the subpar performers are working at Dairy Queen, not for your organization, and when you invite your employees to see how well they did against “the average”, it will only hurt you.
Think you can keep reviews a secret? Think again. Employees share information, or worse, overestimate the grades given to peers. Feelings get hurt, grudges get nursed, productivity suffers.
So what’s a smart manager to do? Frequent, informal, and specific feedback about job performance. Stagger salary reviews so that there’s no one “big-bang” moment for annual performance reviews.
The Absolute Value Problem
Imagine toll-booth operators. People drive up all day, hand the guy a dollar, and drive on. Drive up, pay, drive up, pay…routine and predictable.
What’s the difference in bottom line difference between the best performers and the worst performers in this job? Not very much. The two biggest risks are probably (a) the guy abuses customers so much they choose alternate routes, or (b) theft. The first guy is extremely rare and can be fired, and there’s a whole cottage industry behind “till reconciliation” that discovers the second guy’s shenanigans before too much damage is done. The upside is also pretty limited. What are you going to do if a toll-booth guy is super friendly? Recommend that lane to your friends? I doubt it.
Now think about software developers. What’s the bottom line difference between the best performers and worst performers? I remember hearing, way back when, anecdotal evidence that it may be as high as 100-1. My own experience suggests that, among actively employed programmers, the difference may be as much as 20-1. On the low end, maybe 10 to 1. Whatever the number is, it’s way different than the difference among toll-booth operators.
Is the difference between the best-paid and worst-paid software person in your organization different by a factor of 100? 20? 5? Probably more like a factor of two, maybe three. How about the annual raises coming out of a performance review? Again, probably two or three. No organization I’ve ever heard of would justify a salary increase or bonus of 50% for normal circumstances. So in your annual review, you are actually UNDERPAYING your high performers and OVERPAYING your low performers, relative to each other. And the system is structurally set up such that you, the line manager, have very little effect over what you might call the “Big-Oh” bounds of this calculation.
Software people are talent, not “resources”. They can’t be levelled. They can’t be optimized via sophisticated bean-counter spreadsheets. They can’t be exchanged like cogs in a wheel. They can’t be managed like toll-booth operators. Your organization will suffer if you don’t remember that.