Steve Ballmer and the Microsoft Review
Now that Ballmer is retiring from Microsoft everybody is piling on him for all manner of perceived failings as CEO. Most recently for the toxic culture that the Microsoft review system is said to have engendered within the company.
Now I hate to do this because ordinarily I take great glee in razzing Microsoft for its blunders but in this case, I have to take Ballmer’s side of the argument. Criticizing Microsoft is fun, but having actually been CEO of several technology companies, I’ve got to say I’ve developed some sympathy for Ballmer’s position over the years. One of the biggest challenges I’ve always experienced making big technology is that you’re only as good as your tools and your tools are people. When I have great people working for me… I become a great CEO. When I have less-than-great people working for me… I’m a less-than-great CEO. Which very quickly makes you realize, as a technology company CEO, your primary job is recruiting and retention. I don’t think that a lot of people would dispute that Microsoft has managed to recruit and retain a lot of top notch talent over the years. It’s always been one of Microsoft’s great strengths.
It’s also the case that unlike almost any other industry, software development can have a very extreme productivity curve. There are engineers out there who can out code 5-20 ordinary skilled programmers. There are one in a thousand engineers who can out-innovate nearly all ordinary skilled programmers. If you wanted to design a compensation system that rewarded programmers in direct proportion to their contributions to a products success the contribution curve would generally show that 98% of your people were contributing between -1X to 3X units of productivity relative to their peers and the remaining 2% of employees where contributing 3-20 units. In other words the top 2% of your team might be contributing 20%-40% of a successful products value. In traditional organizations top performers were promoted to management in order to justify their higher compensation but in a technology company converting a great engineer into a bad manager is the last thing you want to do. You need that engineer to stay exactly where he is in the organization but compensate him “fairly” for his productivity. That means you need a review system that can pay one engineer 5X-20X what a peer sitting next to him in the same cube is earning.
On such a review scale the majority 99% of your employees are going to sit in a very tight cluster. Most of them will never have the potential to be 10X contributors and a few are active productivity inhibitors that need to be removed from the organization if they remain that way.
Microsoft’s review system was designed very early on to try to capture these observations about engineering performance. The fact that the Microsoft’s TRIES to pay exceptional people exceptionally is a great thing! Most companies are content to pay all of their employees a flat market salary that sits somewhere near the market average for their position and experience + or – maybe 25%. Microsoft “solves” this problem in a mathematically “obvious” way. The company generates profits which, combined with market pricing for talent is used to set a “budget” for review time. That budget is fixed. It doesn’t matter how many super-talented geniuses worked on a product if the product didn’t perform exceptionally… the value of their productivity may still be low despite their best intentions and hard work. That fixed “review” budget has to be distributed among those team members as closely as possible to each member’s contribution to the products success. On a team of 100 people 2 of them may have written 20% of the used code and solved design or performance problems in an innovative way that nobody else could have discovered. A few of the team members may have worked 80 hour weeks on code that all had to be thrown away in the end because it wasn’t used in the final product. How do you distribute raises fairly?
Under the Microsoft review system the team’s manager would stack rank the team based on a large number of subjective variables. The purpose of the stack rank is to force the manager to honestly assess the value of each team member to their organization. A stack rank prevents a manager from being lazy by reviewing all of his team members to be equally valuable. That stack rank is then used to distribute review scores and compensation along a bell curve proportionate to each team member’s position in the stack. Since the manager has to deliver the reviews, he had better be prepared to defend them to his employees. A few top performers get a lot, a few bottom performers get “the shaft”. If you want a promotion at Microsoft, you need to review near the top of your stack several times. If you review at the bottom of your stack several times you will be given a certain amount of time to find a new job in a different part of the organization… or be terminated.
This review system is the source of endless whining, chaffing and grieving at Microsoft. Almost nobody likes it. It doesn’t try to coddle emotionally fragile people and make them feel good about their mediocre contributions to their team. There are no easy places to hide for the lazy and incompetent. Managers are forced to be clear and accountable to their teams about each members relative value… which most people don’t want to hear if they are not #1. Top performers, who should be happy, often feel guilty that they are making so much more than their less valued friends and team mates. There is social pressure for them to commiserate with their less rewarded peers. It is said that the system results in heavy internal politics and competition, that people on the same team are incentivized to back-stab one another, that managers will deliberately seed their teams with mediocre talent so they can pay their top people and close friends more generously and that Microsoft lost top talent because of it. Etc.. etc.. Although all of this is true to some greater or lesser degree, I don’t believe that any of it is necessarily a BAD thing. Here’s why…
How to “Prove” your review was “unfair”?
Quit and get a better job.I know of very few people who have an accurate perception of their own value to an organization. Do useless incompetent idiots KNOW they are useless incompetent idiots? Isn’t one of the hallmarks of the most useless incompetent idiots, that they zealously believe they are top performers? Do all exceptional people KNOW they are exceptional? The Microsoft review system is a very clever economic structure for correctly valuing people over time regardless of their imagined value… which is almost always greater than their real value. There is a simple way for anybody unhappy with their Microsoft review to PROVE that they are incorrectly valued. They can quit and take a higher paying job at another company. Anybody who remains at Microsoft after a dissatisfactory review and bitches endlessly about it, is just WRONG about their actual value… clearly nobody else in the market is willing to pay more to hire them. Microsoft measures its churn rate to determine how FAIR its compensation system really is. If Microsoft’s overall churn rate is too high then it can INCREASE its review budgets to bring it’s aggregate churn rate into a desired range.
Microsoft WANTS to churn out weak employees. If a manager is deliberately seeding his team with weak employees in order to pay his top contributors MORE, then Microsoft’s system will force that manager to automatically churn them out, forcing the manager to spend more time recruiting WEAK talent from other internal groups. Thus, there is even an economy and some job security for bottom performers. If you believe that this is a widespread practice at Microsoft then imagine how unproductive you must be to not be able to find another job in the organization when you have been forced out of your current one as a result of constantly getting reviewed at the bottom of the stack?
Does the system really incentivize backstabbing within a team? That would tend to assume that the team manager is UNAWARE of this incentive AND that the team is so small that backstabbing a few members has a chance of boosting an employee’s stack rank into the vaunted top 2% stratosphere… otherwise backstabbing, which takes time and effort, would not be a particularly advantageous activity unless the team manager actively rewarded it. Since the overall size of the compensation budget is proportionate to the company’s actual performance, engaging in backstabbing would have the collective effect of reducing the size of everybody’s compensation by making the company less productive.
In short the underlying economic incentive ideas behind Microsoft’s review system are BRILLIANT and provide inherent resistance to fraud and gamesmanship. That does not mean that such a brilliant system can’t be poorly administered or that people will LIKE being judged and compensated fairly. If a manager rewards gamesmanship or tries to curry personal favor with disappointed employees by blaming the system for their review then the system will obviously be less effective. If employees with an inaccurate view of their “worth” believe that getting an accurate assessment of their value to the organization is equivalent to being told they suck every review period then it would also cause an understandable morale problem… among the mediocre… which they can fix by improving their performance by accepting that they can do better or by proving the review wrong by quitting and taking a higher paying job elsewhere. People who remain in the same job and remain disgruntled about their review are just proving that they have, apparently, been assessed accurately… Moreover, that being delusional about their real value and developing a poor attitude when challenged to perform better, are two areas they could improve on.
Now the California school of millennial management thought on this subject is that it is more important to make employees feel happy about their jobs and peer relationships than it is to treat them fairly. We all know that happy employees are more productive don’t we? Also because the market for engineers is so competitive, unhappy, less productive employees can always quit and get another job elsewhere, sometimes with a raise simply because there is such a scarcity of talent. Better to try to keep everyone happy and forgo the whole systematic measurement of productivity and fair compensation for the sake of institutional peace?
Do useless incompetent idiots KNOW they are useless incompetent idiots?
Bill Gates and Steve Ballmer however would tell you that being surrounded by mediocre talent they have to carry and not being recognized for their contributions is precisely what guarantees that the top 2% performers will quit and go elsewhere. Mediocre people can always be replaced. The most valuable employees are hungry, aggressive and embrace the challenge of climbing to the top of a stack ranking rather than making excuses for their shortcomings and developing attitude problems when they don’t get their way. People who consistently sit at the top of their stack ranking will eventually get promoted into a different stack (with higher standards) thus making room for their former peers to move up. Bill and Steve would say that keeping the best and brightest people challenged was more important to retaining them than keeping everybody “happy”.
Of the many failings that Ballmer can be ascribed, Microsoft’s review system is hardly one of them. I believe that he and Gates knew exactly what they were doing when they designed it and that it was responsible from a very early stage for cultivating a culture of excellence among Microsoft’s BEST performers and a culture of systematic and rationale productivity measurement. I believe that Microsoft’s struggles under Ballmer are the result of too much success and complexity for anybody to manage at that scale, which is hardly an indictment of his tenure as CEO.
This is particularly a problem in the Midwest. The notion of paying one employee so much more than another (and possibly more than their managers?) doesn’t fly here. Businesses would be much better off in many cases offering bonuses to it’s current competent employees to achieve some goal that adds value to the business, but we don’t do that. Instead we say we need to fill X seats to launch a new initiative by X day, and we wind up filling them with warm bodies making market salaries who slow down the entire effort.
There’s an old school “manufacturing” mindset that’s still all too common. The notion that tech professionals are essentially Keyboard Operators on an assembly line making a product makes it all too easy to compensate people in a way that is utterly in-congruent with their actual value added. The people who are bothered by this start companies…
Yes, that mindset is found all over. It’s also tragic that our tax system has evolved away from encouraging companies to share the wealth with employees in this manor. Stock options used to be a great vehicle for companies to “bonus” employees in a way that didn’t cost them immediate cash and gave them ownership in the company proportionate to their contributions as well as a way to reward people for retention. Options were also a great way “pay” people more without creating a wide disparity between peers in their immediate income. It was generally perceived that “Seniority” corresponded to greater wealth, not semi-annual performance which was easier for your less “compensated” folks to accept. Microsoft really pioneered this approach to compensation as an alternative to classic pension schemes and cash bonus programs.
Here’s another idiotic analysis piece;
http://qz.com/118513/the-long-hard-road-back-for-microsoft/
“Low morale and a destructive internal culture
What was at the root of Microsoft’s broken product pipeline? As outlined by Vanity Fair’s expose on Microsoft’s lost decade, Ballmer was a tone-deaf manager. Microsoft’s “stack ranking” system of management, in which employees were graded on a curve that meant that one in 10 members of a team always had to receive a rating of “poor” even if everyone in a group was an A player, pitted employees against one another, discouraged collaboration between and even within teams, and slowed Microsoft’s development process to a crawl.
”
This was ALSO the system under Gates… and who believed that everybody on their team was an A player? The team manager writing the reviews did that. If it was TRUE and A players were getting bottom stacked ranked then they were leaving those groups and taking jobs in other groups where they WERE the top performers, thereby ensuring that the companies best talent was distributed relatively uniformly. If they WERE NOT actually all A players that information was also revealed when a supposed A player changed teams and still remained at the bottom of the stack. All organizations, even ones composed of all A players, experience churn, Microsoft’s system was also effective at helping to ensure that the natural churn rate tended to happen among the bottom performers NOT among the top ones. Why would you leave a group in which you were consistently at the top of the stack?
The idea that the review system “CAUSED” petty politics and internal competition is also a funny one. First, causing internal competition was a DESIRABLE feature of this system and second pampered type-A hyper-nerds hardly needed an excuse to engage in petty politics, poor morale and unproductivity… Again the California school of Millennial management would contend that reviewing and paying everybody the same would make them happy, more team-spirited and more productive? Yeah… that’s how people actually behave… I love the sweeping condemnation of one of the worlds most successful companies compensation systems without any assertion as to what would magically work better… Microsoft should adopt one of the more standard review and compensation systems widely used by LESS successful organizations? Geniuses…
More moronic analysis;
with-steve-ballmers-departure-a-look-at-microsofts-flawed-system-of-performance-reviews
It’s interesting how nobody ever points out that a review system that requires strict accountability from management and individuals for their performance also tells you a lot about the actual quality and maturity of everybody participating in the system. It REVEALS people who are prone to sulking, lacking initiative, politicking, poor teamwork or other low maturity and integrity behaviors. It’s tragic that our culture produces more and more of them. Is there some review system at other successful companies that PREVENTS people from behaving in these ways? How would they know if it did? What’s their excuse when this human BS goes on in their organizations anyway but they have no tools to quantify it and identify it’s sources? THIS WAS THE SAME SYSTEM IN PLACE during Microsoft’s fastest growth years… now it sucks? idiots…
What large organizations need to function well is high integrity people who are inherently self-motivated and take pride in everything they do regardless of their “environment”. These are the people who don’t require a lot of management overhead and supervision. People who NEED to be artificially motivated by their managers, who NEED coddling and validation, who NEED constant supervision are inherently less valuable and place a higher management burden on the organization. Microsoft’s system REVEALS these people and churns them OUT causing a net accumulation and uniform institutional distribution of self-driven, self-motivating, accountable people. BRILLIANT! Total compensation is set such that the net churn rate is held at an acceptable level and if it is working “perfectly” you would expect the organization to have a high level of dissatisfaction with the system but NOT to the extent that people were willing or ABLE to quit and get higher paying jobs as a consequence in large numbers. If everybody LOVED the system it would mean that Microsoft was PAYING TO MUCH for it’s talent.
Myself and the people I worked with were ALWAYS unhappy with our compensation. We were ALWAYS unhappy about our reviews even when they were exemplary. It didn’t matter to our performance, we loved what we did and we took tremendous personal pride in our jobs without anybody having to “manage” us to do so. We were always pissed off and jealous that there were Microsoft millionaire’s all working around us that we felt had done less to earn their wealth than we did. No matter how many options we were granted at our reviews we were indignant about not getting MORE. After we left Microsoft as millionaires we were still hungry to prove ourselves and started companies of our own. That’s EXACTLY the culture Bill and Ballmer wanted to cultivate and it was a GREAT one for people who really were exceptional.
A great article by one of Microsoft’s great managers and a mentor, John Ludwig.
http://theludwigs.com/2013/08/i-have-a-different-view-on-stack-ranking-than-the-spate-of-critical-microsoft-articles/
What’s interesting here is that you see bitter Microsoft employees blaming the review system for their recession experience. Microsoft tightened up and paid less during a devastating recession and instead of being grateful for having jobs, they’re PISSED at the review system. If it was so unfair… why stay? oh the economy is a total wreck and nobody will pay you more than Microsoft… hmmmm…
What’s comical about reading this is that they sound exactly like the people in my day who were bitter because they were becoming millionaires slower than their peers. It was just SO UNFAIR AND DEMOTIVATING! I remember when I started at Microsoft being told that I had joined too late to get rich and my stock options were worthless and how unfair the review system was then because of it. (The stock split 5 times while I was there)
A good alternative perspective to what I have read about stack ranking elsewhere.
Perhaps the key point is your ending: “Microsoft’s struggles under Ballmer are the result of too much success and complexity for anybody to manage at that scale”.
MIcrosoft is not a company simply churning out billions of widgets. It probably has at least 20,000 software engineers. And half of those are doubtlessly egotistical prima donnas. How do you manage that?
Maybe these megacorps just can’t survive thru purely internal efforts. Much of Cisco’s growth in past decades has been thru acquisitions. MIcrosoft has also done their share of those. Even Windows owes its roots to the acquisition of QDOS. Most large companies are constantly doing acquisitions and divestitures.
It’s a tough one. Anytime I see somebody say “Ballmer should just do X” it’s clear that they don’t understand that Ballmer can’t make a herd of cats do anything specific, nobody can. It would add real intellectual value to the dialog if someone brilliant suggested a better way to increase the probability of getting a desirable outcome from a herd of cats. Obviously I have some thoughts on it but I’ve never managed a mess that big, I had my hands full with a few hundred cats.
I don’t think acquisitions often work out well anymore in the SOFTWARE tech space. The problem is talent churn, it’s just so high that buying a technology company is like trying to buy a hole in a lake. Very often it’s just gone with a “bloop” and you’re out a ton of money. how do you retain and motivate people that KNOW they are rich now and just biding their time to quit? Skype appears to be a pretty smart acquisition for Microsoft but their record on that front is maybe 20% hits. The bits are only as valuable as the people who know how to use them.
My first instinct would be to look for ways to simplify. I know something that always slowed Microsoft innovation down was preserving one giant consistent monolithic platform, shipping DirectX as a library allowed us to circumvent that in the early days. I’d be inclined to start more small skunk works teams with more freedom and HIGH rewards for wild success. Then look at systematically folding the “hits” back into the platform. The problem with that “brilliant” idea is that it doesn’t leverage Microsoft’s platform strengths unless internal groups can come up with ideas that span Microsoft’s products and channels. That suggests that you need an incentive for other product teams and divisions to “adopt” skunk works ideas from other groups they see potential in because doing so and helping them succeed leads to outsized rewards.
I hate to say it but my concession to the dictatorial approach would be to look at more stringent approaches to ensuring that consumer experience of ALL Microsoft products is highly respected. Sloppy user experiences just do a ton of damage to the MSFT brand. It really didn’t take a massive media launch to “Discover” that the Windows 8 UI was a terrible idea or the XBOX ONE message was a train wreck, those were easy disasters to avoid. I think that’s a MAJOR area where Apple has been beating the heck out of them. They need a much higher bar of accountability for their consumer experiences and messaging.
Late to this party, but I consulted in the Windows CE area at Microsoft in 1998-99. A couple of notes:
1) The people working on the SEGA Dreamcast made changes to the kernel without telling anyone else and those then became the core of the system. There was no coordination or design to the process. The memory management design was an externalization of the SH4 memory map. It was lunacy. But it happened because the various teams were in conflict with each other.
2) A guy wrote a new implementation of the registry for Windows CE and, at least while I was there, it was not documented. He added features that he wanted. And it was shipped. He did it because the people who were the keepers of the registry code for desktop OS would not cooperate with him.
3) I named a certain behavior a “meeting torpedo” and when I explain it to former Microsoft employees, they erupt in recognition. A meeting torpedo is a group program manager or higher who slides down the hallways and when he sees a meeting to which he was not invited, slips in and blows the meeting up. The meeting would effectively end when the topedo would start ranting about schedules.
I went down to Apple for an interview while working at Microsoft and the atmosphere could not have been more different. In Redmond (at least in building 32) people shuffled around, never looking up, never making eye contact. In Cupertino, people were happy and seemed to like each other.