Hacker News Reader

Comment by silverbax88 | Hacker News Reader

"Eichenwald’s conversations reveal that a management system known as “stack ranking”—a program that forces every unit to declare a certain percentage of employees as top performers, good performers, average, and poor—effectively crippled Microsoft’s ability to innovate."

This is the most important sentence in the article. Because many companies over the years - a lot of them start ups - would point to this practice by Microsoft and try to follow their lead. Microsoft has been stack ranking almost since conception. It was always a horrible way to run a company. But it took decades for it to actually bring the company down.

View on HN

Novell did something similar. They used to fire IIRC the 'bottom' 10% every year. This HBS article talks about the system: http://hbr.org/2001/05/leading-through-rough-times-an-interv... Interestingly, Eric Schmidt seems to have modified it a bit and seen that it wasn't working well.

"Incidentally, not all systemic changes work. I’ve also learned that certain management techniques can actually make things worse when applied to a distressed culture. For example, I had always worked in companies with yearly and quarterly employee ranking systems, in which people were divided into three groups: overperforming, performing, and underperforming. So not long after I came here, we started a ranking program that graded on a curve: 45% into the overperforming group, 50% into the performing group, and 5% into the underperforming group. I didn’t know—and certainly didn’t intend it this way—that if you got the lowest grade, it was presumed that you were about to be fired. We started getting hate mail from people who argued that there was no way to rank people who worked as a team. The ranking system exacerbated the culture of fear and proved to be such a huge retention and motivation issue that we were forced to stop it after a year. In its place, we introduced a modified ranking program that better reflected overall employee performance."

It seems like one factor in that scheme that would imply you're going to fire "underperformer"s is that 95% of people can't be classified that way. The names are symmetric; why not something more like 20% - 60% - 20%?

The other factor would be that there's an idea (which I can't speak to personally) that the only purpose of performance evaluations is to protect the company in case a fired employee sues to reinstate their job, which would mean that an "underperforming" grade was a necessary prereq to firing.

The overall flaw in this system is that it assumes that there are always people worth firing. But some companies - and let's think of smaller start ups as an example - try to hire really well, and there are some shops that are loaded with talented people. Stacking them and firing the 'worst' performers because of a metric is basically just cutting talent. My take is that if a company thinks they have 10% of people on staff worth firing every year, fix the hiring process - that's the problem.

Firing the bottom 10% company-wide is quite different from firing the bottom 10% from each department, regardless of the performance of that department relative to others, which is closer to what the article describes for Microsoft. As the original article noted, being in a 10-person "unit" meant that 1 out of those ten was guaranteed a bad review, even if that unit as a whole somehow generated 50% of Microsoft's profits for the quarter.

The article is completely wrong on this point. In 99% of cases, people are stacked in a sufficiently large group that everyone on your immediate team could hypothetically get a great review.

While that's true, it is also reliant on the fact that the manager of your immediate team be able to effectively argue for your team members during the stack-ranking meeting. If you happen to have a manager that ends up being too reticent, many of the underperforming scores could end up being disproportionately assigned to your team.

You're definitely right on that. However, if your manager isn't willing or able to effectively argue for your team members, it doesn't matter what the review system is: you'll still get screwed.

Read the article closely it says "Every current and former Microsoft employee I interviewed—every one—cited stack ranking as the most destructive process inside of Microsoft, something that drove out untold numbers of employees". Are you saying the employees were lying or that the interviewee misheard? Its not possible to argue against this point.

Also, what you say about a sufficiently large group never happens. What actually happens is each lower level manager is forced to rank his or her team members along the curve so as to make the process "fair". And inevitably, this gives rise to a large host of harmful behaviors - slacking off on work by people who have low self esteem as they will get ranked low "anyways", grandstanding, people trying to take credit for others work, infighting, corporate politics, people waking up and starting to work hard closer to the appraisal cycle to capitalize on the "recency effect", dysfunctional teams that deliver only because of heroic efforts from a couple of the team members.

Any system can be gamed this one has got gamed out long ago, my experience - I have seen it do nothing but harm in every single case.

That was an idea they (probably) stole from Jack Welsh. http://en.wikipedia.org/wiki/Vitality_curve

I was employed in a company that used something like stack ranking. It was brutal having to evaluate the dev team I was leading. There were a couple veteran developers that consistently did excellent work, but there was a rookie that was learning incredibly fast and producing very high quality work.

On a scale of 1-5, I felt all three of them were 4.5, but I couldn't submit scores like that because of stack ranking. The question then became whether to piss off one of the veterans or to discourage the rookie by giving a < 4 score.

Yup, stack ranking effectively punishes good teams. Why should a manager strive to consistently hire the best people possible when some of them are going to have to be unfairly ranked?

Hmm, so the ranking system would actually encourage leaders of good teams to have hire poor performing members to act as their sacrificial employees come review time.

I've heard stories of exactly this happening. People hired just so they could absorb negative review scores. This is, obviously, dysfunctional on many levels.

Something similar happened to me. It's not necessarily that the hire is intended to be sacrificial, but it sometimes falls to the "new guy" to take the hit.

So after about 3-4 months of working at Microsoft, I was ranked in the bottom 10% (the bucket for people who don't have good prospects for furthering their career at Microsoft.) This was back when they used a 10/70/20 system.

Now, don't get me wrong, I don't think I'm some master developer that deserves to always be in the top 20%. I strive to be above average, but I consider myself to be merely average, especially compared to many of the exceptionally bright people working there.

Here's the real bitch. The reason I was given for this rating had nothing to do with my performance. I shit you not, and this is as verbatim as I remember it: "Somebody has to be in the bottom bucket and since you're new, I can't rank you higher."

The problem is that this stuff fucks with your career there. It's a black mark on your record that follows you around. HR assured me that someone being new is not a valid reason to rank someone like that, you have to base it on actual performance.

An easy fix that my previous company had employed: have a "too new to rate" category.

In retrospect, I just had a shitty boss, and it wasn't necessarily reflective of Microsoft as a whole. I also should have challenged it more vigorously, but I didn't want to create animosity between my boss and myself seeing as I was barred from transferring for.. 18 months was it?

I'm not too bitter about it though; I ultimately ended up resigning after a year and then coming back to work for the same department as a contractor at a much higher salary. As such, I no longer had to participate in performance reviews at all.

I ultimately ended up resigning after a year and then coming back to work for the same department as a contractor at a much higher salary

While that might have worked out OK for you. It's a beautiful example of this kind of dysfunction.

Avoiding nonsensical peer reviews (never seen one that made any sense) is one of the reasons I went contracting.

My company does stack ranking, and my first performance review mirrors yours almost to a T. And the distrust that it has created between my direct manager and his reports is horrendous. He always prefaces his evaluations (discussed personally with each of us) with "I'm limited to giving Exceeds Standards to only two members of the group." Said group having 18 members. He's not required to grade any employees as Does Not Meet Standards, unless they truly do. But know that I'm lumped in at the exact same level as the remaining sixteen coworkers is extremely disheartening. This just encourages employees to quickly determine the firing line.

The minimum performance required to avoid termination.

Well, it's that or have them on welfare. :)

This blew my mind. I was thinking it, but after reading your words it hit home.

The more prevalent variant on this is managers would leave "room" in their review model for themselves. Come in under the curve and you leave room that makes it that much easier for your manager to reward you.

Not exactly. It does encourage a manager to increase the 'beta' of his team. That is, adjust hiring standards to take more risks and filter out the subsequent higher flow of poor performers in exchange for finding (but not necessarily retaining) exceptional performers.

stack ranking actively hinders managers from retaining exceptional performers. if you only have 2 A grades to give out and you have five people deserving of it ALL of them are going to be motivated to switch to lower performing teams or entirely new employers.

any sort of 'objective' evaluation of employees is prone to being gamed and any gaming of evaluation is actively harmful to corporate culture/performance. witness google where lower profile or troubled projects are abandoned en masse by any employees with the ability to turn them around

At that point, I would just drop the person who I dislike the most to the bottom of the bucket. After all, as a manager, I have no choice but to fill that slot, and given every-one has performed well on the job, what other heuristic can I use to select the "bottom 2"?

Or I may drop the person who I can get rid of with the least hassle, maybe the new guy who just joined the team or someone who has taken long leave during this period.

See how it works? Nothing to do with performance, and inevitably its the "A" performers who quit in disgust. And companies wonder why they cannot get talent to stay...

Equally worse is giving points to your team based on the budget you have to increase salaries. A system I currently live under. the result is stack ranking stuck at the 70% part... Death to motivation if you ask me.

They used this as the point for why Microsoft has fallen behind. The former developer says, "I was too concerned with appeasing my manager's when I should have been concerned with being a better engineer."

But really? Is that why Microsoft has fallen by the wayside (realtively)? Because they don't have good enough engineers?

I disagree. I think they have a wealth of talented engineers.

What they need(ed) are people with vision, a deep understanding of the future of consumer electronics and computing, and a way to execute on that vision. They need to deliver short, medium and long-term products that feed their vision. They need to show people the future of the industry instead of lagging behind it.

Engineering is a crucial component in that, but it's only part of the story IMHO.

I disagree. I think they have a wealth of talented engineers. What they need(ed) are people with vision

Vision doesn't happen in a vacuum. It happens within the wider cultural context of the organisation. A technical lead, a project lead, a manager - they are all affected by the environment they are in. If you spend all your time fighting politics and trying to shore up demoralised engineers, you are not going to waste your time dreaming about ponies and rainbows.

If you want an analogy, you can take the world's greatest general, but if you drop him in a WW-I trench, he is just cannon fodder.

"If you spend all your time fighting politics and trying to shore up demoralised engineers, you are not going to waste your time dreaming about ponies and rainbows."

I think that's a fair point. Good reply.

"I disagree. I think they have a wealth of talented engineers."

That statement is false in many ways. As a developer, you keep seeing substandard products from Microsoft. Have you looked at the architecture of SharePoint? compare SqlServer CE to SQLite, not to mention the horrible tutorials and source code which Microsoft keeps putting out.

C#/.NET is an excellent platform/language but Microsoft keeps putting out horrible sub standard code/tutorials

Have you seen the sales of SharePoint?

Engineers always think that companies are engineering driven, when in fact mostly they are sales driven. If SharePoint sells well, then its not substandard, even if you don't like its architecture.

Sales? I thought it was just a free add on?

Your first two paragraphs are not mutually exclusive. My former employer had a wealth of talented engineers who were overridden or ignored on key architecture decisions, either by semi-technical upper management or customers. Things had to be done a certain way because X wants them done that way, and by the time the bad decisions get highlighted the technical debt is too high for management to justify changing direction, and the program is forced to limp along with the crap they have.

My impression of Microsoft (entirely from the outside) is that they have this same problem.

In a large company, the technical documentation may be scrubbed by non-engineers.

It's a mix. I've seen some brilliant engineers here but I've seen a lot of so-so employees collecting pay checks. What I do think is very real issue and more so than the stack ranking is that we are incredibly date sensitive here and cost sensitive. I often feel, in my personal opinion, that improvements or changes that could make us better off 13 months down the road are culled or punted because the upfront cost is too high or because set commitments for an upcoming timeline.

What is the point of having all those billions in cash on hand, and an R & D budget exceeding the market cap of many DJIA participants, if you're going to penny pinch where it counts? http://finance.yahoo.com/q/ks?s=msft What timeline is so important (short term) if you lose the future, period?

Look at Sharepoint's history. It's a Frankenstein of a bunch of acquired products. None of the built in services were enterprise ready last I checked (e.g. an alarm that doesn't always fire, a file upload service that doesn't always upload, etc., etc.).

I have to agree.

The article started with stack ranking, but clearly there were much bigger issues looming.

>“He didn’t like the user interface, because it didn’t look like Windows,” a programmer involved in the project recalls... “Windows was the god—everything had to work with Windows,” Stone tells Eichenwald. “Ideas about mobile computing with a user experience that was cleaner than with a P.C. were deemed unimportant by a few powerful people in that division, and they managed to kill the effort.”

They could have been years ahead on phones, tablets, and e-readers. But instead they had a culture that tried to shoehorn a standard Windows experience into every device (I imagine for marketing reasons). They are only now getting over it a decade later, though it's probably more accurate to say they have the same philosophy but in reverse: shoehorn a mobile experience into the desktop in Windows 8.

What difference does it make if you have the greatest engineers in the world but management is blocking them at every turn?

"shoehorn a mobile experience into the desktop in Windows 8."

I actually laughed out loud when I read that--it's true!

I think most of the comments here are missing a key ( and in my opinion the most detrimental) effect of stack ranking at microsoft: it prevents talent from clustering around good projects (unless they are so good as to be exempted from it).

Because someone always ends up on the bottom, a fair number of devs don't want to end up on who already has a top quality dev. Worse yet, in order to keep good people, managers don't want / will promise not to bring in more good people. Thus there is an active incentive to recruit( at least internally) mediocre talent since a team needs to hire people to keep headcount and needs to pad the middle of it's stack ranking.

Yeah, in theory this shouldn't happen because the rankings for each team are merge and those on the bottom of a good team should come in above the bottom of a bad team. In practice that seems not to work because both ICs and managers seem to have the above philosophy.

In education we call this 'norm referencing'. The highest scoring 10% get A, the next 15% down get a B and so on. In the UK, people have moved to 'criterion referencing' so certain performance criteria are set and if you reach those, then you get that grade.

Would the second practice, given sensible and negotiated targets work better?

Both Accenture and UBS do the same thing. I never liked it, myself, for the reasons cited in the article. This type of performance system means it's far more important to be good at schmoozing than to be good at working.

I worked at Andersen Consulting (which is now Accenture) for long enough to be inside the banding process as well as subject to it and actually thought that the ranking system worked pretty well -- in a consulting company.

To begin with, it didn't work the way described in the article (does Microsoft's system actually work that way? if so -- ugh). There were five bands, and the "you get fired" band didn't generally get anyone assigned to it. Furthermore, the percentages in each band were somewhat flexible and applied across the organization, not to any one group of people. It was quite possible for an effective team to all be band 1 and 2. (But then, AC didn't really have fixed teams.)

Of the four non "you're fired" bands, 1 was excellent, 2 good, 3 adequate, 4 was improve or leave. (I don't think we were even required to assign any band 4s.)

But a consulting company is very different from a product company. A consulting company actually WANTS a large proportion of its workforce to leave within a certain time because ultimately it's a pyramid scheme. Only 10% of people who join get to associate partner (or whatever) level and you need to advance at a certain rate or you'll be ticked off, so you need to shed 90% of people before they've lasted X years or you're screwed.

A product company needs people with deep technical skills to hang around. It's one thing to use "up or out" to drive your sales force (and the upper echelons of a consulting company are salesmen) and quite another to do this to your deep technical people.

So it's a misapplication of a management theory and not an intrinsically bad management theory.

Agreed that the model makes a lot more sense for a consulting company as opposed to a product company.

However, in terms of how it actually worked, I was also involved in meetings to allocate people to bands, and the percentages had to be respected across the client group (UBS, in that case).

This means that if there were, say, 10 analysts in the UBS client account, then only a couple could get "Exceptional" (or whatever it was called), and so on...

The implication of this was that your performance (and therefore pay raise, promotion, and so on) largely depended on whether:

a) you had made a strong enough impression on your manager so that he/she would be motivated to do a decent job of fighting for you in the meeting;

b) your manager was actually politically competent enough, and had the political capital to spend on getting you a good rating;

c) you had made enough of an impression on all the other people likely to be at the meeting so that they'd know about it and accept it when your manager suggested that you should be rated Exceptional.

If you hit all three of those, you might get a great rating. If you didn't bother to network with everyone who was likely to be at the meeting, or if your manager sucked or didn't like you all that much, you were pretty much fucked.

I agree with pretty much everything you've said, but the point is that AC was (Accenture is) a consulting company which is all about selling services and impression management (and seldom about the substance of the work, which is frequently copy-paste-and-edit stuff pulled out of Lotus Notes with the previous client's name and particulars replaced with the new client's), so the reward structure strongly mirrors what's good for the organization. And forcing people -- often excellent people -- out of the organization is actually a _beneficial_ side effect. (Yes, this is a very cynical view of management consulting, but I think I have some credibility here.)

The problem with Microsoft -- assuming we're talking about an accurate representation of their system -- is that the reward structure rewards visibility rather than results, but the organization actually wants results.

Coming up on my first performance review at Accenture. I'll see how flawed the system is in a few weeks ;).

The problem is with what the viable alternative is for companies that are huge like this. It's just tough. Maybe we all just need to stop worrying about how other humans measure our performance/value.

Good luck!

Definitely don't "stop worrying about it" - or you'll get screwed.

I disagree. Stop working as a permanent employee for big companies. Then you'll get paid more money and not have to do these stupid performance reviews. If you consult for a company that tries to do performance reviews, run because you've hit the "bad company" full house.

So when I was at Google, hiring was not done by the groups with the headcount. Instead they were hired centrally and then later dispatched. The reasoning for this is that for a manager, someone who is just good enough will generate utility and thus should be hired. The central hiring is supposed to enforce companywide higher levels.

I wonder if stack ranking is the same but for getting bad people to leave? Even a weak employee can be better than no employee for a manager...

This seems like a good place to mention the original stack-ranking critic, Mini-Microsoft (still going!):

http://minimsft.blogspot.com/2004/07/microsoft-stack-rank-as...

Valve uses stack ranking as well. Does Valve do anything different which makes them immune to the same company-ruining internal competition?

It's probably not competition that causes stack ranking to be a bad idea -- it's the measures that determine the ranking. Valve's flat hierarchy probably means that you only need to earn the respect of your peers, not of a group of managers that determine the rankings. I could see a situation in which a truly engineering-driven culture combined with stack ranking would become a meritocracy that rewarded the best engineers.

(a) Valve does it company wide

(b) As I understood it, it is only used to determine comp adjustments

If you go into it with the view "We're all excellent people, and these people are the best of an excellent bunch, so they're getting the biggest raises this year", it seems like a reasonable system.

Seems like it could appear to work in a growing, young company with rising stock value. If someone is genuinely good, but gets stack-ranked out of your group, it's probably not as hard to move to another, possibly new group, when the firm is growing.

But when the company is mature, and the stock is flat, and the headcount is under pressure...

The problem all these companies face is how to determine raises fairly. To do that, you can allow each manager to determine his employees' raises independently, which typically means the manager will exaggerate their performance, or you can devise some sort of system based on numerical reviews. But how do you create such a system? Every one I've ever heard of has been unfair and political in some way - but still probably less unfair and political than each manager picking numbers for their own team.

I'm trying to say this is a hard problem, and picking on the downsides of "stack ranking" is easy if you don't consider alternatives.

This is a fairly common practice across many relatively successful organizations. In addition to the other companies mentioned in this thread,I believe McKinsey consulting also takes a similar approach.

Granted, successful does not necessarily imply they are innovative. This management strategy may be more effective in more sedate and established industries where adhering to best practices may be more important than charting new courses.

If anything, this would seem to be a cautionary tale to tech startups that may be tempted to hastily implement traditional MBA'esque management styles. One size certainly does not fit all.

IBM does something similar as well.

I believe GE does also. If these kind of companies are employing the same system with better results, would this imply that the theory behind stack ranking is not the problem?

GE's system (at least the one originally set up by Jack Welch) isn't quite the same, although the idea is similar‡.

Basically there are 3 "bands" in the Welch system (with the lowest band being 10%). You reward the top band (20%) heavily, the middle band (70%) modestly, and dump the bottom band (10%) every year.

In some ways, the intent is similar to the economic class system (incentivize the middle class to become the upper class).

GE had a lot of success when they adopted that model, although obviously, they did a lot of other things as well which may have been a more direct effect (getting rid of a lot of businesses they were in, for example).

I can appreciate the ideal of constantly cutting out the bottom-performing resources, provided that the mechanism to track performance is sufficiently quantifiable (and accurate). I haven't heard enough from people at GE as to whether or not that is the case there, though.

‡ http://en.wikipedia.org/wiki/Vitality_curve#Straight_from_th...

Practically, I think these systems suffer terribly from inertia effects. The top stays the top, the middle stays the middle, because it causes so much angst to move someone from the top to the middle. So the middle just gets a job somewhere else.

GE's executive team claims otherwise[1]. "The top 20% and middle 70% are not permanent labels. People move between them all the time. However, the bottom 10%, in our experience, tend to remain there."

[1]: http://www.ge.com/annual00/letter/page4.html

No. It is just that those companies are too big to have collapsed from it yet.

Or they run the system with non-fixed bands, where I guess it could be made to work.

Reply on news.ycombinator.com