Beyond stack rank: How Microsoft needs to change now

In the startup world, I’m prone to saying, “You can take the boy out of Microsoft, but not the Microsoft out of the boy.” This week I’ve been tracking the controversy around Microsoft’s abolition of the old stack-rank review system, and couldn’t help but have very strong feelings on the matter despite having left the company back in 2004. I am also seeing a lot of strong feelings from our GeekWire readers, including people who are currently at Microsoft or have recently left. As chairman and co-founder of GeekWire, I think this is awesome. Discourse in our community on this subject is great.

But as a former Microsoft manager, I think the issue is really complicated.

My own 12 years at Microsoft were entirely under the “old old” review system, using a 2.5 to 5.0 scale, where stack-ranking exercises were a common practice amongst managers. In fact, we used to refer to the broader stack-ranking exercises, when teams would merge their respective lists, as ‘horse trading.’ I hated these horse-trading meetings. I would’ve rather gotten a root canal while getting my legs waxed. They took time away from building products.

In my opinion, the most challenging aspects of the old system were:

• You could have a uniformly high functioning team of rockstars but *someone* had to be in the middle and bottom. This sucked because some very good people got told they weren’t good enough.
• Knowing this, people often over-functioned in unsubtle ways before every review cycle to make sure their hard work was known, and credit received. Presumably this is one of the things that the new system is seeking to address.
• Merger exercises with other team managers consisted of you arguing for your people being higher in the larger merged stack, at the expense of others whom you did not manage or know very well. Managers were, in essence, making data-blind arguments about others’ careers.

In the words of one GeekWire reader: “At worst it fostered lying about your direct reports’ qualities; at best it forced managers to scour and find something, anything that could be used as a mark against hard-working employees.”

So it was absolutely not a perfect system. Absolutely not. But there were *some* advantages.

• You always knew who was critical to your business, and who was less so. I think this is a best practice of any business leader (startup founder or division head within a big company). You should always know who you can’t live without. There have been some unfortunate cycles of layoffs at parts of Microsoft due to business contraction, and at those times you simply need to know which assets (tech and personnel) are necessary to keep and which are not.
• The system taught new managers how to make very tough tradeoffs. The old review system taught at least this young manager that sometimes the business outcome of something can feel in conflict with the social or organizational outcome. In its most abstract form this is just simply an important business and life skill.
• Gosh darnit, you simply can’t promote everyone, or give everyone the same raise every review period. You have to make financial tradeoffs. As another GeekWire reader put it: “Now we don’t have to tell 50% of the people they performed below the mean of performance … but everyone gets the same financial result anyway? This may make people feel better … but it’s hard to see how it will drive better individual performance. Now it will be much easier for managers to say ‘everyone’s great’ and ignore the mathematical fact that half of the people are not above average.”

Fair points, all. But in the final analysis, If I had to choose, I’d choose the new system, which will let managers hand out raises and bonuses at their discretion, within the limits of their overall budget for compensation.

The bottom-line is this; no review ‘system’ comprised of human beings is going to be perfect. But at least giving managers more discretion and power to allocate reward and compensation incentives within their teams will give them more leeway to manage teams in accordance with their particular business circumstances, and to do so in a management style that befits the manager or culture of the team.

Moreover, change is good. I absolutely applaud the company’s ability and willingness to change. Change is what this company needs right now. One of the biggest management challenges at Microsoft has always been the constant cat-herding of super-high-functioning Type A employees to align to a single mission (be that a product vision, or even just a process for product creation). It can take 20 minutes for a group of partner level engineers to agree where to have lunch. Thus, I cannot imagine how supremely difficult it must have been for Lisa Brummel to have spearheaded this fundamental change.

Whether we think it’s too late, too soon, not enough, or too much … I’m not sure that it really matters long term. The key is that change happened, and in a specific vector to improve both morale and collaboration. Both things are precisely what the company now needs. Microsoft finally gets to A/B test the new solution vs. the old. If this new system is wrong, then by all means they should ‘fail fast’ and figure out an even better solution.

But lastly I’ll add, a change in a review process alone isn’t going to change culture. How do I know this? After Microsoft I actually worked for another large, successful, publicly traded tech company with a rigorous stack ranking system, but yet colleagues there were uniformly collaborative, collegial, communicative, and just dang nice. And yes, every quarter you’d sit in a room with other managers and the division VP to stack rank super-smart folks and make very tough tradeoffs.

This company was called Google, and they created products that were just as good. Perhaps the difference was that the workforce at Google was 20 yrs younger, more earnest, and didn’t have big mortgages to worry about. Perhaps it’s simply the well-documented differences between how millennials were raised vs. us genX-ers and boomers. Googlers have also never suffered a major drop in the stock price like Microsoft did, avoiding the fear that can create in the workforce. My point isn’t about Google vs. Microsoft, but about how review systems and culture are more independent, than dependent.

I suspect every current manager at Microsoft knows this by now, but a change to a more collaborative and less abrasive culture can only come about when individuals refactor their work styles authentically. This is the stuff that’s baked into the DNA of the people you hire, and may not be changeable overnight. It’s trusting each other, and being honest. It’s giving colleagues the benefit of the doubt. It’s avoiding going behind each other’s backs. It’s communicating openly and dealing with conflict early and accurately. It’s erring on the side of over-communicating with others, etc. The sole reliance on a systemic fix takes ownership of personal behavior away from the worker.

So in addition to changing how it reviews employees, Microsoft should also re-train some of them. And even further, Microsoft will also need to change how it recruits new ones.

And in all these regards, I sincerely and earnestly wish folks at Microsoft the best of luck. I’m rootin’ for you. All your alums are rootin’ for you. Your shareholders are rootin’ for you. The tech community is rootin’ for you. Let’s do this!

Jonathan Sposato worked at Microsoft from 1992 to 2004. He sold his startup Phatbits to Google in 2005, and then later sold another startup, Picnik, to Google in 2010. He is currently the chairman of GeekWire, PicMonkey, and Vizify, all companies that do not stack rank.