This kind of stack ranking was used at large investment bank (starts with M) about 10 years ago while I was working there as a C++ developer.
It resulted in a strategy where some teams have hired puffer people with the only purpose to have poor performance ratings assigned to them and let them go when the demand for such inevitably rose for such from HR.
This shielded the core team allowing them to focus solely on actual business requirements.
There were people in the office taking decent salary, doing barely anything with kind of no expectations of their performance whatsoever. Some were doing their masters, some hit the gym very hard, some were just chilling out.
This was my first lesson on how focus on tactical goals and metrics can fundamentally screw up the long term vision.
Extremely well put.
Any time a business decides on a metric / decision criteria, they introduce a strong incentive for their employees and managers to act accordingly.
For some reason, leaders rarely seem to think hard enough about the possible outcomes.
My anecdote - at a former big tech company, we had a long hiring freeze, and they even decided that when someone leaves or gets fired, their headcount goes back up to senior leadership who decide where that headcount should be (so as a manager, if you fire someone - you might lose the position entirely).
So, naturally, even since then there have been zero people fired, even if they are low performers, since managers want to keep their headcount. But talented people obviously still get offers and leave, and when they leave, their headcount disappears.
So after a year or so of this great policy, the average level of talent across the company is steadily dropping, which also means it'll be much harder to recruit top talent even after this policy is over - because most talented people want to join a team where they can still learn and evolve, collaborate with other smart folks, etc.
I have had several conversations where I explained that the consequence of a policy decision would be negative, and then the policy was rolled out anyway.
For example, I explained that if it takes a foot-high pile of paperwork to obtain an internal use SSL certificate, then the highly predictable outcome of that policy is not "greater security", but unencrypted HTTP everywhere.
The inevitable response to such feedback is that "people ought to do the right thing", which is laughable on its face. Nobody, and I mean nobody in their right mind will subject themselves to months of paperwork, procedure, and pleading for something that they consider to be optional.
"But, but, the right thing is for them to jump through those hoops anyway!" is the typical response to such logic.
This kind of argument is just futile. Someone up high has decided that they know better, that human nature will bend to their will, and will forge ahead despite being told they are wrong.
> leaders rarely seem to think hard enough about the possible outcomes.
This really annoys me about politics. They seems to reason as: We have problem A, so let's do solution B to solve this. And then the reasoning ends.
What should happen is to evaluate all the consequences of solution B. And sometimes those consequences work in such a way that they make problem A even bigger.
Reasoning about an interconnected system is crucial here. And as developers we are very well aware of systems and the impact changes can have. But it seems a lot of people are not able to reason about systems.
> Extremely well put. Any time a business decides on a metric / decision criteria, they introduce a strong incentive for their employees and managers to act accordingly. For some reason, leaders rarely seem to think hard enough about the possible outcomes.
I’m so tired of this advice. Maybe metrics are bad, but the only thing worse than metrics is no metrics.
Discouraging an action just because it doesn’t work 100% of the time doesn’t seem efficient either.
The argument really should be that certain metrics are bad. Sure, propose a better one, but saying that leaders don’t think hard enough just because a metric turns out to have ill effects seems useless advice. How can you be certain in advance when you have thought hard enough?
A better approach is probably to recognize that metrics are hard. Keep an eye on them, iterate, and be willing to make mistakes as that’s the only way to improve.
>For some reason, leaders rarely seem to think hard enough about the possible outcomes.
I think it's the same issue politicians have. No one of those in power ever thinks of unintended consequences. My guess: it takes a special kind of anal person to actually want to get into power (and thus to eventually get into power) - and they just can't imagine people acting not according to their great master plan.
> So, naturally, even since then there have been zero people fired, even if they are low performers, since managers want to keep their headcount. But talented people obviously still get offers and leave, and when they leave, their headcount disappears.
This sounds like a former professor of mine who would order phones and tables for everyone from the spare money on the end of the fiscal year for "research purposes" since otherwise the department would lose the budget.
Agree. Happened at the IB I worked at. I remember a particular person who I was just baffled they kept around, so I asked his manager. "He's my designated fire-ee" I was told. "If in a given year I have to make a cut, he's the guy I cut and I know it won't affect productivity in the team. If someone leaves organically then he survives for another year". A lesson in perverse incentives.
As far as I know, stack ranking was first used, and with good success, by law firms. The thing is, the people being evaluated that way were on a career track to become partners in the firm, and as such they were evaluated by peers or rather future peers. Doing that with artists, writers, programmers and testers who have no expectation of eventually becoming board members and major shareholders is a whole different story and makes no sense.
So long as you (the employee) know the score this sounds like a sweet gig. Work for a year or so at a nice leasuirely pace, save up, get to put large investment bank on your CV, move on.
It’s the employees who don’t know that they’be been hired as the designated sacrificial goat I feel sorry for: That’s a potentially awful position to be in.
This sounds like a (horrible) way to implement universal basic income. Just hire puffer people, have them do nothing. They get fired, hire them somewhere else, rinse, repeat.
Except it's extremely demotivating for people doing actual work and getting paid similar money.
My first job was in an old newspaper printing business (this was back when newspapers were highly profitable and a respected medium). We had a few old people in our department who were basically kept there doing nothing because they didn't want to retrain on new technology but they were friends with management so they weren't let go before retirement. Which sounds super humane and socially sensitive. Except there were a few of us new hires that couldn't get permanent contracts because the departments were "fully staffed", we were busting our ass and making 1/2 of what these guys were for doing nothing. And this "fair treatment" wasn't there for everyone - just the people who were friends with upper management - basically if you paid tribute by inviting a few people up to the management offices now and then, bring a bottle of nice booze and locking the door for a few hours, doing out of work favors, etc. There were people who were laid off and who had to retrain to remain.
It was a sobering experience after my rosy childhood world views about fairness in society.
UBI is mostly so people have a minimum, don't starve, and can live in something that passes as a dwelling. Not having basic housing is a huge poverty trap, so make sure folks can eat and be kinda stable.
These guys are puffers at big investment banks getting well over 6 figures while they pump iron and work on a masters degree. No comparison, or sympathy.
Perhaps apocryphally (I can't find a valid source at the moment), Keynes was supposed to have suggested during the Depression that people be hired to dig holes and then fill them back up.
Next time you need a puffer person, please let me know.
Would be awesome to get a 6 figure salary on the side while I work on my startup.
(I'm very good at leetcode-interview BS so will have no problem passing your interview and will be happy to take the severance when you need to fire me)
Hiring managers: Don't fall for the parent's game. Their leetcode skills prove that they are just trying to trick you into hiring them, after which they will become a valuable member of your team, making your next firing more difficult. If you really want a puffer person, hire me. I'll take the position with no interview, demonstrating my commitment to putting zero effort into your project.
Edit: And don't worry about any bidding war with other managers. I'll just put you in my queue and rotate through you all as needed.
> There were people in the office taking decent salary, doing barely anything with kind of no expectations of their performance whatsoever. Some were doing their masters, some hit the gym very hard, some were just chilling out.
I am pretty sure I am this person right now. Got the job but mostly not qualified for it, though I'll concede it's a niche field and they probably have to train people regardless. But it feels square-peg round-hole.
Always keep your CV up to date (its soo hard to write a CV the day after you get fired), browse the job ads and watch for the buzz words and keep the skills in those words fresh and go to the gym.
(I actually think most stack ranking is dumb, but the comment you made is a very common response and i honestly don't get it)
"This was my first lesson on how focus on tactical goals and metrics can fundamentally screw up the long term vision."
I don't get how you take that away from this - this is a lesson in how bad managers who refuse to do what is asked, and apparently have no accountability, can screw up the system.
Stack ranking may be good or bad, but if your managers are working around the system, and nobody is holding them accountable for it, that's a separate, and honestly bigger, problem.
If they actually did what was really being asked (instead of trying to work around it), either
1. it would work, and the company would be better off.
2. it would fail miserably, and the company would either stop doing that (and be better off) or go bankrupt (and we'd all be better off because it would be an object lesson).
The reason this went badly is precisely because of a lack of real accountability, not because of stack ranking itself.
The point is that if you can game the system, and the system cannot determine that it is being gamed, then it's a bad system.
Who watches the watchmen? In a large-enough organization, the stack ranking system is set by some executive high up the mountain. The low-level line managers and group managers are incentivized to work together to get the budget (for puffer employees) to help them keep their proven performers. Finance doesn't care as long as (a) the bottom-line improvements from firing puffers keep getting realized and (b) management keeps being able to pay puffers below-market salaries. Puffers themselves shouldn't care in this day and age, taking the opportunity to take the puffer paycheck as a 100% remote position with no actual responsibility, while working full-time somewhere else that actually appreciates them, so puffers basically get a free paycheck out of it for nothing.
It doesn't matter if it's a good system in theory. It only matters if it works in practice. There's an old joke about economists - they treat the real world as a special case.
I can't tell if you're trolling or not. What's so confusing about rational actors reacting rationally to a change in incentives? Do you think there's only one way for a policy to "fail miserably" or something? Once someone knows that they'll likely have to fire some number of their team whether they want to or not, of course they'll hire sacrificial lambs. It's not like the upper management that has that stupid of a policy is going to give two shits if some middle manager claims they missed their own performance expectations because of instability in their team.
And think about the corollary: how much additional effort would have to go into monitoring managers to make sure they weren't circumventing the "spirit" of the given bullshit metric? A company like that would resemble a dystopian regime.
> It resulted in a strategy where some teams have hired puffer people with the only purpose to have poor performance ratings assigned to them and let them go when the demand for such inevitably rose for such from HR.
Those people are known as "sacrificial lambs." I ran across the term in a previous position. More than one person specifically included this concept in their planning.
I think there are also two different "flavors" of stack ranking usually at play.
There is the one where the company wants to reward top performers. I think this one is most common, and I have never worked at a company that doesn't so some form of bonus budgeting that tries to capture a group of employees who are performing well and give them a bonus. The budget itself dictates how many employees can fit into that group, so ultimately what happens is a natural bell curve of the biggest rewards going to the perceived top performers, as agreed upon by managers usually at the organization level.
The other more insidious stack ranking involves a punitive element, whereby a group must necessarily be singled out for performance improvement reasons. This is the type of thing where managers, no matter what the team members actually accomplished, have to put someone on a performance improvement plan or start performance managing them to "get them up the bar" because someone naturally has to fall into the "bottom performer" bucket of stack ranking.
I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available. The latter approach is what causes problems with companies, because all of a sudden team mates become competitors and cultures can go downhill rapidly. Think Amazon, Microsoft under Ballmer, and GE under Walsh.
It's kind of interesting how many companies who are doing the former bonus budget pooling eventually get seduced into doing the latter and destroying their culture.
I've seen this play out at my university in two different ways.
The more obvious way is very obvious: we had a professor who told us he'll be grading us by first tallying up the answers, and then applying a bell curve to the scores, so that X% of the students get an A, Y% get a B, etc., and some Z% of the students will fail. As you can imagine, everyone was up in arms about how obviously unfair this is.
The less obvious case, but close to what you describe, was scolarships. For almost every "track"[0], scolarships were awarded for top X% of students. In the "track" I was in, and another one in our faculty, the scolarships were instead given to anyone who crossed a certain absolute grade threshold, in (IIRC three) tiers. The higher the tier you reached, the more money you got, independent of how other students fared (probably subject to budget constraints, though).
Now, I can't prove it, but there was one other notable difference between our "track" and that other one with same scolarship model, and every other "track" at the university I was aware of: ours were famously cooperative, everyone else was fiercely competitive. I don't think it's a coincidence. My university experience was that of everyone helping each other out, because it was always a win-win both socially sometimes financially. Everywhere else, this was unthinkable - helping other students created a risk of them getting ahead of you and taking your scolarship.
I'm dismayed to see the same thing repeating in companies too.
--
[0] - I can't seem to find the correct word to describe the set of students across the years who pursue a degree in a particular subject (e.g. "Computer Science") on a specific faculty. I.e. the vertical slice.
[0] - I can't seem to find the correct word to describe the set of students across the years who pursue a degree in a particular subject (e.g. "Computer Science") on a specific faculty. I.e. the vertical slice.
Many in the US would call it the Department of Computer Science. At my university this was within the College of Engineering, Computer Science & Technology. There was a dozen or so departments in this college, Electrical Engineering, Mechanical Engineering, Civil Engineering, etc. There were other colleges like Arts & Letters, Health and Human Services, etc.
> The higher the tier you reached, the more money you got, independent of how other students fared (probably subject to budget constraints, though).
It worked the same way at my University with the same outcome as yours. 7 years after graduation we still maintain regular contact as a group with my classmates.
'Batch' is the term commonly used for the set of students who start a program at the same time. For instance, the CS MS batch of 2023, who are expected to finish their MS in CS in 2023 (So they all should have stared in 2021 assuming a two-year program).
> I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available.
The problem with this is not the size of the bonus pool.
The immediate problem is the thresholding. Due to the sharp threshold, the bonus will practically never stand in proportion to performance. An employee that is judged to perform at a level of 6.1/7 gets nothing, but the one at 6.3/7 gets a lot. This makes the scheme anti-collaborative.
The less obvious problem is the ridiculousness of the idea that you can assign accurate individual performance numbers to people. These things are insanely noisy to the point of being meaningless. It ends up effectively a lottery. That would be fine if you called it a lottery, but by pretending it's not you create perverse incentives.
There are only two ways to do bonuses well:
- either forego individual bonuses and reward the entire company equally (or maybe scaled by salary) based on company performance, or
- hand out individual bonuses only to true (statistically verified) outperformers -- in this case you might end up not handing out any bonuses most years.
"These things are insanely noisy to the point of being meaningless."
Goldstein and Spiegelhalter did a paper on league tables for organisations (e.g. schools and hospitals) along the lines that noise in the data made the results questionable.[1]
Perhaps the idea transfers to individuals within an organisation that uses stack ranking?
They tend to behave exactly the same though. If you rank only good performers, it's a small step before you use that against bad performers: "you've been with us 5 years and never managed to get on the top performers list. I am afraid you are not a good fit for our culture here"
It's very easy to see how people might make this logical jump, but honestly if your bosses manage you out for never being a top performer they are doing you a favor: you don't have to work for their dumb asses anymore.
Good managers recognize the value of role players. Not everyone can be a top performer. It's an absurd expectation to have for any organization.
I believe evaluating software development only works at the bottom line of tasks with formal metrics like issues, bugs or tickets solved. Also in general only possible in larger teams and probably more for maintenance tasks than active development.
There are no boni for top performers in my company at all. If business is good everyone gets a bonus. The only exception here is sales. Here people are more competitive. I guess I would get martyred if I even suggested it for other departments. And I believe it would be pretty bad for morale and teamwork overall. It is good that people don't use energy to present themselves in a good light and instead solve the problems we have. I still believe there is a lot of incentive for employees to see the company performing well. As you said, I have the impression that this becomes a background issue if employees compete with each other.
There was a time I would love to have worked at Blizzard, even while being aware of the volatility and risk of the industry. I wouldn't want that today regardless of compensation. Still, I don't believe it is either or. Both systems you suggest probably have an overall net negative effect for development. There are other forms for employees to participate on company success. Administration for employee boni is wasteful overhead for the most part.
If it is true what the article reports, I believe the manager did the correct thing. On that metric he might be a high performer among his peers.
I've heard that at companies that did stack ranking, managers of groups with high performers would actually seek out poor performers who didn't care about their ranking from other groups, to add to their group so that they wouldn't have to give a poor mark to their own people. These poor performers became hot commodities. Talk about gaming the system...
I've seen this happen. The best part is when attrition is bad enough in some cycle that the poor performers stick around... and now the managers are really in trouble!
It can be difficult for line managers to fire their entire team.
> These poor performers became hot commodities. Talk about gaming the system...
Couple iterations of that and you'll have people who are hired and/or paid extra for the purpose of being a "fuse" that protects the high performers during layoffs by being first to get fired.
Circa 2006 I saw a friend who was a manager do exactly this. There wasn't stack ranking. He just kept a low performer on his team to sacrifice to the RIFs when they came through.
The bad thing is that this is predictable behavior and management still does it. Should the suggestion get grouped into the 5% of low performers? This is clearly damaging to overall work morale and company goals. And not to an insignificant part.
What would you give for that performance? I cannot see anything that would benefit the company. Not even employment costs, because employees certainly price that in.
> A Blizzard spokesperson told Bloomberg that the evaluation process is designed to [...] "ensure employees who don't meet performance expectations receive [...] differentiated compensation.
Most of the response is the usual PR speak, but "differentiated compensation" seems like a new level of back-bending to avoid saying "less pay."
This blurs the real issue. You can still have different compensation and PIPs and everything else without stack rankings.
The real issue is that stack rankings compel managers to meet a quota for low performers. Even if the team is full of top tier engineers, one or more get dealt the short straw every review period.
While I'm just a developer, my wife is a director at the same company.
I multiple years, she's turned in her evaluations and had the director and VP above her tell her to lower several employees ranking.
One example was an engineer who didn't get along with the people on his team for a pretty high profile project. They know the guy's brilliant and don't want to fire him, so they move him to a lower profile project to decommission several server farms they had been struggling to get done. Dude goes in and absolutely kills it. They estimate he saved the company hundreds of millions of dollars by the tens of thousands of servers he decommissioned and renegotiated several large contracts the company was on the hook for.
My wife ranked him in the upper tier. He wrinkled a few feathers, but once they reallocated him, he absolutely thrived and saved the company literally a ton of money.
Nope.
He pissed too many people off, her boss and bosses boss mandated she give him a low ranking with a miniscule pay increase and minimum bonus in order to punish him for his behavior early in the year. Nothing he did after that mattered to them. Nothing.
When your work is a popularity contest and not measured on actual performance, its a toxic mix.
I wonder if people pushing for stack ranking are justifying this in terms of natural selection. That is, even if the company is all top performers, periodically cutting bottom 1% or 5% or whatever is going to improve average performance even further over time.
This would make some sense if employees were like cells or fruit flies. But in most companies, employees are like organs. You can't improve a body by cutting out 5% of its organs and waiting for them to reappear. At best, you'll degrade the body's performance. At worst, the body will die.
It's also funny how they try to wrap it so that it almost sounds like they're doing it for the benefit of the employee:
> "ensure employees who don't meet performance expectations receive more honest feedback, differentiated compensation, and a plan on how best to improve their own performance."
Honest feedback? Great! A plan for how to improve my performance? Sure, sounds good [if you don't know what the social realities of a PIP are]. Differentiated compensation? Wait a second...
I don’t know about you but far before I showed up for the very first day of my career my expectations were that if I performed better than my peers I would be paid more and if I underperformed I would be paid less. And generally that has proven to be roughly true, with the occasional weird stuff here and there.
The only place I know were this does not seem to be the case are government employees were everybody is paid according to a fixed grid and were salary increases are purely time gated. My experience with working with people compensated on a such a model has been… interesting.
If your compensation consists of base pay plus a bonus, that is usually tied to performance, then why should you get the bonus payout if you don't meet expectations?
There are employee evaluation tools that try to measure if employees are meeting expectations.
The entire point of stack ranking is that it does not do this. The article is about a lead who resigned/got fired because the system wanted to penalise someone who was meetings expectations.
There's nothing wrong with getting paid less if you don't meet performance expectations. I'm just impressed with the term they made up ("differentiated compensation") to try to make that sound nicer.
Companies still use it. The company I retired from three years ago still uses it. If the manager is doing their job, their team’s performance, relative to the rest of the company, is not normally distributed. Stack ranking is a stupid idea, as many of Welch’s ideas are now considered.
Curve fitting "performance" to an "expected" bell curve is clearly idiotic, even from a mathematical perspective. In the described case one won't have a symetric bell curve and the population is going to be too small, even if measuring performance was done right, which it clearly wasn't. Whoever invented this process has done so without understanding normal distributions. It's a bunch of mumbo jumbo.
Doesn't a normal distribution typically arise during random sampling? I.e., if there's some statistic you calculate based on a sample, then when you take a bunch of independent population samples, that statistic will tend to be normally distributed over those samples.
If that's the only reason we'd typically expect to see a normal distribution, then I don't understand why we'd expect employee ratings to be normally distributed.
Why is it distributed? Maybe if you took people at random but:
(sarcasm) shouldn't you have all good people because your leetcode hiring process only produces "the best" (/sarcasm)
Add to the fact that "rankings" are often largely political. There's a "minimum talent" aspect and then the last 1-2 points is most often political and representative of machiavellian power levels in the corporate dominance hierarchy.
Also, the article headline should explicitly call out Stack Ranking.
Stack Ranking basically admits one of two things:
1) your recruiting sucks
2) your work culture sucks
Well, and #3: corporate culture doesn't value balance or having any "bad times" in your life.
> The process, called "stack ranking," requires managers to give about 5% of their employees a low performance evaluation to fit on a bell curve of relative performance.
...companies still use it ? Thought microsoft taught industry a lesson here
I managed a small team in a company that then used stack ranking. I hated the process and as others here mentioned pointed out many times applying a bell curve to a small population is bad math. While we still used the process I saw everything from voting blocs, managers intentionally submitting all of their employees ranked up one notch from where they should be, managers collecting "dirt" on employees managed by others so they could shoot down those employees during the ranking process and more.
The process itself was flawed because people had a tendency to stay in the rank where they were submitted because frankly managers didn't have the time, energy or interest in really talking through every employee. It was just easier to rubber-stamp the whole process and move on. Also it was nearly impossible to get rid of the process because of course people who had succeeded in it didn't want change. I couldn't in good conscience continue like that, so I took over rankings for my location. It went from one day to a week-long process and I made people vote anonymously and incorporated a ranked choice system and then aggregated the results. It was a far more time consuming process but IMHO fairer. I still hate forced ranking despite succeeding in it. It becomes part of your job - managing to the metric.
Vestiges of it remain because people still don't know how to distribute compensation without ranking people, particularly for employees that have the same review rating.
The problem with the 2000s-era Microsoft method, as I understand it, was that the population sizes they used were too small. Stack ranking was done at the individual team level, and that did indeed lead to people sabotaging each other, because in a team of 10 only 2 people could get the highest pay raises and 1 person would be placed on a PIP (or whatever the ratios were).
But yes, all major companies still do something like this AFAIK, just with a larger population size now. Speaking from my experience at Amazon, the target was around 4%, but this was across entire departments. My department was 300 people. Is it possible that you can have a department of 300 people where you can't find 12 people who are below-average performers? Yes. Is it likely? No.
I don't know how many people the lead in the article managed. If it was a team of 15, he has a legitimate gripe. If it was a team of 200, not so much.
That’s one of the two but the bigger problem is how you measure performance. It’s notoriously hard to measure outside of the simplest jobs and once you tell people they’ll be punished unless they juice a particular number, well, that encourages all kinds of undesirable behaviour: rigging tickets or other metrics, encouraging people not to do maintenance programming or hard tasks, setting competitors up to fail, not doing things to help or train less experienced colleagues because that isn’t captured, etc.
Yes, managers are supposed to keep that from happening but this policy tells them that a) you don’t trust them to do their job, b) their job now focuses on protecting their people from the system, and c) being busy playing the same game on their own personal behalf.
>Is it possible that you can have a department of 300 people where you can't find 12 people who are below-average performers? Yes. Is it likely? No.
I mean, below average performers will be around 40-60% just because how averages work.
The problem is that the question is "is their performance low compared to team", while the question you should be answering is "is their performance low compared to the market.
If team's management is on point with hiring the "worst" won't be that much worse than average anyway, so if you want to replace the bottom few % you essentially waste time as new hires might be at same productivity level.
On other side if you recruiting sucks and 20% of the dept does the 80% of the job, it isn't going to help either.
No manager has direct insight into performance of 300 people (at least in tech). They have to rely on subordinate managers/teamleads, each of which can be unreliable for different reasons. Some may be playing politics, some may just be bad managers. Some will overhype their employees, some will be honest to a fault. Get them all in a ranking board, and all sorts of games get played.
If your team leads could be trusted to measure appropriately, you wouldn't need stack rankings because they'd correctly identify those under-performing. Stack ranking is conceding that your managers can't properly measure performance. Shouldn't that reflect more on management?
The stack ranking on my team at MS was done across a team closer to ~60 developers. Each manager managed 5 or 6 and then they would all get together and fight for their team members. It was insanely toxic. Not sure if every team was the same.
If the facts don't fit your theory, change the facts! The US obsession with force-ranking everything is unhealthy, and I say that as someone who has often been fortunate enough to end up on the right hand tail.
>> Stack ranking has received continuous criticism since General Electric popularized it in the 1980s.
I didn't know this existed until now. The problem with this (among others) is that I can never personally know where I stand. If I'm doing well, but being excelled by others, I still get penalized. There's no guarantee, it's almost random.
Also, I now have a new question to ask in interviews, and if it does exist in a hiring company, I can negotiate by declining bonus pay performance reviews and in lieu of ask for higher wages (not that any company would go for this).
Ah the 1980’s when Neutron Jack Welch (he kills the people but leaves the buildings intact) was seen as a visionary. He was ahead of the curve in massive layoffs while doing tons of mergers and acquisitions that years later seem to have been questionable at best. Nice to see this MBA hookum continue to survive like some radioactive zombie cockroach
That's about the only time I can think of where it makes sense. If you have to lay off some people, then obviously you want to get rid of the lowest performers.
But if times are good and everyone is performing well enough and contributing, why get rid of them? If they're not performing well enough, then you shouldn't need stack ranking to figure that out, and you don't need to arbitrarily set a percentage of employees to put on performance improvement plans or fire.
Just move somewhere people are incompetent and you'll be safe and the best programmmer. Unfortunately when you hit a new crossroad, you'll need to reach out to outside peers to get moving in the right direction and spend a lot more time on self study.
I am refusing to believe that a company has less than 5% low performant employee ratio. With pretty much everyone is pulling their weight with no employees that are plain slacking and/or bad hires that delivers less than expected.
In fact 5% is quite a low estimate in my opinion. I would estimate it 20%, easily 1 in 5
Maybe there is a bigfoot out there but I will refuse to believe until I see one
Stack ranking seems to be a solution for managers that are not doing their job. Either because they are trying to be nice or oblivious or just don't care. Since this guy quited to me it sounds like the first.
Back in the early 2010s, EA and Ubisoft were considered the worst gaming studios because they pushed "loot crate" mechanics. Now, Activision-Blizzard has not only developed some of the most expensive stores in any game, but they almost always deliver half-finished games and abuse their employees in the process, while EA and Ubisoft are considered some of the best places to work in the gaming industry. I think this company needs a boycott really bad to put it back on the good path.
> EA and Ubisoft are considered some of the best places to work in the gaming industry
Oh, really? I haven't followed this kind of stuff in a while. So they're good now?
How would they rank for software development, in general? Game studios are notoriously subpar when you start comparing them to FAANG, big enterprise sotware development shops, etc.
> Game studios are notoriously subpar when you start comparing them to FAANG, big enterprise sotware development shops, etc.
I don't think this has changed substantially. It's supply and demand: lots of people want to build games for the sake of building games. At least in theory it sounds more exciting than maintaining whatever microservice for a FAANG. So, because there's a lot more supply than demand, they get paid less.
Unisoft is an absolute shitfest internally. EA on other hand was always a good place to work, despite being an plague of the gaming industry on itself. Doing good things and being a good place to work are not necessary the same thing.
No, they just started paying PR firms when their deserved reputation started causing brand new graduates (their core hire) to pass on their job offers. The entire game industry is a massive shit show of employee abuse.
Readit News
It resulted in a strategy where some teams have hired puffer people with the only purpose to have poor performance ratings assigned to them and let them go when the demand for such inevitably rose for such from HR. This shielded the core team allowing them to focus solely on actual business requirements.
There were people in the office taking decent salary, doing barely anything with kind of no expectations of their performance whatsoever. Some were doing their masters, some hit the gym very hard, some were just chilling out.
This was my first lesson on how focus on tactical goals and metrics can fundamentally screw up the long term vision.
My anecdote - at a former big tech company, we had a long hiring freeze, and they even decided that when someone leaves or gets fired, their headcount goes back up to senior leadership who decide where that headcount should be (so as a manager, if you fire someone - you might lose the position entirely).
So, naturally, even since then there have been zero people fired, even if they are low performers, since managers want to keep their headcount. But talented people obviously still get offers and leave, and when they leave, their headcount disappears.
So after a year or so of this great policy, the average level of talent across the company is steadily dropping, which also means it'll be much harder to recruit top talent even after this policy is over - because most talented people want to join a team where they can still learn and evolve, collaborate with other smart folks, etc.
For example, I explained that if it takes a foot-high pile of paperwork to obtain an internal use SSL certificate, then the highly predictable outcome of that policy is not "greater security", but unencrypted HTTP everywhere.
The inevitable response to such feedback is that "people ought to do the right thing", which is laughable on its face. Nobody, and I mean nobody in their right mind will subject themselves to months of paperwork, procedure, and pleading for something that they consider to be optional.
"But, but, the right thing is for them to jump through those hoops anyway!" is the typical response to such logic.
This kind of argument is just futile. Someone up high has decided that they know better, that human nature will bend to their will, and will forge ahead despite being told they are wrong.
This really annoys me about politics. They seems to reason as: We have problem A, so let's do solution B to solve this. And then the reasoning ends.
What should happen is to evaluate all the consequences of solution B. And sometimes those consequences work in such a way that they make problem A even bigger.
Reasoning about an interconnected system is crucial here. And as developers we are very well aware of systems and the impact changes can have. But it seems a lot of people are not able to reason about systems.
I’m so tired of this advice. Maybe metrics are bad, but the only thing worse than metrics is no metrics.
Discouraging an action just because it doesn’t work 100% of the time doesn’t seem efficient either.
The argument really should be that certain metrics are bad. Sure, propose a better one, but saying that leaders don’t think hard enough just because a metric turns out to have ill effects seems useless advice. How can you be certain in advance when you have thought hard enough?
A better approach is probably to recognize that metrics are hard. Keep an eye on them, iterate, and be willing to make mistakes as that’s the only way to improve.
I think it's the same issue politicians have. No one of those in power ever thinks of unintended consequences. My guess: it takes a special kind of anal person to actually want to get into power (and thus to eventually get into power) - and they just can't imagine people acting not according to their great master plan.
This sounds like a former professor of mine who would order phones and tables for everyone from the spare money on the end of the fiscal year for "research purposes" since otherwise the department would lose the budget.
It’s the employees who don’t know that they’be been hired as the designated sacrificial goat I feel sorry for: That’s a potentially awful position to be in.
My first job was in an old newspaper printing business (this was back when newspapers were highly profitable and a respected medium). We had a few old people in our department who were basically kept there doing nothing because they didn't want to retrain on new technology but they were friends with management so they weren't let go before retirement. Which sounds super humane and socially sensitive. Except there were a few of us new hires that couldn't get permanent contracts because the departments were "fully staffed", we were busting our ass and making 1/2 of what these guys were for doing nothing. And this "fair treatment" wasn't there for everyone - just the people who were friends with upper management - basically if you paid tribute by inviting a few people up to the management offices now and then, bring a bottle of nice booze and locking the door for a few hours, doing out of work favors, etc. There were people who were laid off and who had to retrain to remain.
It was a sobering experience after my rosy childhood world views about fairness in society.
These guys are puffers at big investment banks getting well over 6 figures while they pump iron and work on a masters degree. No comparison, or sympathy.
According to OP's description it is neither universal nor basic.
Spoiler: it didn't end well.
Would be awesome to get a 6 figure salary on the side while I work on my startup.
(I'm very good at leetcode-interview BS so will have no problem passing your interview and will be happy to take the severance when you need to fire me)
Edit: And don't worry about any bidding war with other managers. I'll just put you in my queue and rotate through you all as needed.
https://en.wikipedia.org/wiki/Goodhart%27s_law
https://en.wikipedia.org/wiki/Perverse_incentive
So ... how do I apply for such a puffer position?
I am pretty sure I am this person right now. Got the job but mostly not qualified for it, though I'll concede it's a niche field and they probably have to train people regardless. But it feels square-peg round-hole.
What do?
"This was my first lesson on how focus on tactical goals and metrics can fundamentally screw up the long term vision."
I don't get how you take that away from this - this is a lesson in how bad managers who refuse to do what is asked, and apparently have no accountability, can screw up the system.
Stack ranking may be good or bad, but if your managers are working around the system, and nobody is holding them accountable for it, that's a separate, and honestly bigger, problem.
If they actually did what was really being asked (instead of trying to work around it), either
1. it would work, and the company would be better off.
2. it would fail miserably, and the company would either stop doing that (and be better off) or go bankrupt (and we'd all be better off because it would be an object lesson).
The reason this went badly is precisely because of a lack of real accountability, not because of stack ranking itself.
Who watches the watchmen? In a large-enough organization, the stack ranking system is set by some executive high up the mountain. The low-level line managers and group managers are incentivized to work together to get the budget (for puffer employees) to help them keep their proven performers. Finance doesn't care as long as (a) the bottom-line improvements from firing puffers keep getting realized and (b) management keeps being able to pay puffers below-market salaries. Puffers themselves shouldn't care in this day and age, taking the opportunity to take the puffer paycheck as a 100% remote position with no actual responsibility, while working full-time somewhere else that actually appreciates them, so puffers basically get a free paycheck out of it for nothing.
It doesn't matter if it's a good system in theory. It only matters if it works in practice. There's an old joke about economists - they treat the real world as a special case.
And think about the corollary: how much additional effort would have to go into monitoring managers to make sure they weren't circumventing the "spirit" of the given bullshit metric? A company like that would resemble a dystopian regime.
Those people are known as "sacrificial lambs." I ran across the term in a previous position. More than one person specifically included this concept in their planning.
There is the one where the company wants to reward top performers. I think this one is most common, and I have never worked at a company that doesn't so some form of bonus budgeting that tries to capture a group of employees who are performing well and give them a bonus. The budget itself dictates how many employees can fit into that group, so ultimately what happens is a natural bell curve of the biggest rewards going to the perceived top performers, as agreed upon by managers usually at the organization level.
The other more insidious stack ranking involves a punitive element, whereby a group must necessarily be singled out for performance improvement reasons. This is the type of thing where managers, no matter what the team members actually accomplished, have to put someone on a performance improvement plan or start performance managing them to "get them up the bar" because someone naturally has to fall into the "bottom performer" bucket of stack ranking.
I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available. The latter approach is what causes problems with companies, because all of a sudden team mates become competitors and cultures can go downhill rapidly. Think Amazon, Microsoft under Ballmer, and GE under Walsh.
It's kind of interesting how many companies who are doing the former bonus budget pooling eventually get seduced into doing the latter and destroying their culture.
The more obvious way is very obvious: we had a professor who told us he'll be grading us by first tallying up the answers, and then applying a bell curve to the scores, so that X% of the students get an A, Y% get a B, etc., and some Z% of the students will fail. As you can imagine, everyone was up in arms about how obviously unfair this is.
The less obvious case, but close to what you describe, was scolarships. For almost every "track"[0], scolarships were awarded for top X% of students. In the "track" I was in, and another one in our faculty, the scolarships were instead given to anyone who crossed a certain absolute grade threshold, in (IIRC three) tiers. The higher the tier you reached, the more money you got, independent of how other students fared (probably subject to budget constraints, though).
Now, I can't prove it, but there was one other notable difference between our "track" and that other one with same scolarship model, and every other "track" at the university I was aware of: ours were famously cooperative, everyone else was fiercely competitive. I don't think it's a coincidence. My university experience was that of everyone helping each other out, because it was always a win-win both socially sometimes financially. Everywhere else, this was unthinkable - helping other students created a risk of them getting ahead of you and taking your scolarship.
I'm dismayed to see the same thing repeating in companies too.
--
[0] - I can't seem to find the correct word to describe the set of students across the years who pursue a degree in a particular subject (e.g. "Computer Science") on a specific faculty. I.e. the vertical slice.
It worked the same way at my University with the same outcome as yours. 7 years after graduation we still maintain regular contact as a group with my classmates.
The problem with this is not the size of the bonus pool.
The immediate problem is the thresholding. Due to the sharp threshold, the bonus will practically never stand in proportion to performance. An employee that is judged to perform at a level of 6.1/7 gets nothing, but the one at 6.3/7 gets a lot. This makes the scheme anti-collaborative.
The less obvious problem is the ridiculousness of the idea that you can assign accurate individual performance numbers to people. These things are insanely noisy to the point of being meaningless. It ends up effectively a lottery. That would be fine if you called it a lottery, but by pretending it's not you create perverse incentives.
There are only two ways to do bonuses well:
- either forego individual bonuses and reward the entire company equally (or maybe scaled by salary) based on company performance, or
- hand out individual bonuses only to true (statistically verified) outperformers -- in this case you might end up not handing out any bonuses most years.
Goldstein and Spiegelhalter did a paper on league tables for organisations (e.g. schools and hospitals) along the lines that noise in the data made the results questionable.[1]
Perhaps the idea transfers to individuals within an organisation that uses stack ranking?
[1] https://www.bristol.ac.uk/media-library/sites/cmm/migrated/d...
Good managers recognize the value of role players. Not everyone can be a top performer. It's an absurd expectation to have for any organization.
There are no boni for top performers in my company at all. If business is good everyone gets a bonus. The only exception here is sales. Here people are more competitive. I guess I would get martyred if I even suggested it for other departments. And I believe it would be pretty bad for morale and teamwork overall. It is good that people don't use energy to present themselves in a good light and instead solve the problems we have. I still believe there is a lot of incentive for employees to see the company performing well. As you said, I have the impression that this becomes a background issue if employees compete with each other.
There was a time I would love to have worked at Blizzard, even while being aware of the volatility and risk of the industry. I wouldn't want that today regardless of compensation. Still, I don't believe it is either or. Both systems you suggest probably have an overall net negative effect for development. There are other forms for employees to participate on company success. Administration for employee boni is wasteful overhead for the most part.
If it is true what the article reports, I believe the manager did the correct thing. On that metric he might be a high performer among his peers.
It can be difficult for line managers to fire their entire team.
Couple iterations of that and you'll have people who are hired and/or paid extra for the purpose of being a "fuse" that protects the high performers during layoffs by being first to get fired.
1: https://www.inc.com/jason-aten/amazons-controversial-hire-to...
What would you give for that performance? I cannot see anything that would benefit the company. Not even employment costs, because employees certainly price that in.
Seen this happen for teams where the manager loves everyone on their team.
Most of the response is the usual PR speak, but "differentiated compensation" seems like a new level of back-bending to avoid saying "less pay."
The real issue is that stack rankings compel managers to meet a quota for low performers. Even if the team is full of top tier engineers, one or more get dealt the short straw every review period.
While I'm just a developer, my wife is a director at the same company.
I multiple years, she's turned in her evaluations and had the director and VP above her tell her to lower several employees ranking.
One example was an engineer who didn't get along with the people on his team for a pretty high profile project. They know the guy's brilliant and don't want to fire him, so they move him to a lower profile project to decommission several server farms they had been struggling to get done. Dude goes in and absolutely kills it. They estimate he saved the company hundreds of millions of dollars by the tens of thousands of servers he decommissioned and renegotiated several large contracts the company was on the hook for.
My wife ranked him in the upper tier. He wrinkled a few feathers, but once they reallocated him, he absolutely thrived and saved the company literally a ton of money.
Nope.
He pissed too many people off, her boss and bosses boss mandated she give him a low ranking with a miniscule pay increase and minimum bonus in order to punish him for his behavior early in the year. Nothing he did after that mattered to them. Nothing.
When your work is a popularity contest and not measured on actual performance, its a toxic mix.
This would make some sense if employees were like cells or fruit flies. But in most companies, employees are like organs. You can't improve a body by cutting out 5% of its organs and waiting for them to reappear. At best, you'll degrade the body's performance. At worst, the body will die.
> "ensure employees who don't meet performance expectations receive more honest feedback, differentiated compensation, and a plan on how best to improve their own performance."
Honest feedback? Great! A plan for how to improve my performance? Sure, sounds good [if you don't know what the social realities of a PIP are]. Differentiated compensation? Wait a second...
The only place I know were this does not seem to be the case are government employees were everybody is paid according to a fixed grid and were salary increases are purely time gated. My experience with working with people compensated on a such a model has been… interesting.
The entire point of stack ranking is that it does not do this. The article is about a lead who resigned/got fired because the system wanted to penalise someone who was meetings expectations.
Doesn't a normal distribution typically arise during random sampling? I.e., if there's some statistic you calculate based on a sample, then when you take a bunch of independent population samples, that statistic will tend to be normally distributed over those samples.
If that's the only reason we'd typically expect to see a normal distribution, then I don't understand why we'd expect employee ratings to be normally distributed.
(sarcasm) shouldn't you have all good people because your leetcode hiring process only produces "the best" (/sarcasm)
Add to the fact that "rankings" are often largely political. There's a "minimum talent" aspect and then the last 1-2 points is most often political and representative of machiavellian power levels in the corporate dominance hierarchy.
Also, the article headline should explicitly call out Stack Ranking.
Stack Ranking basically admits one of two things:
1) your recruiting sucks
2) your work culture sucks
Well, and #3: corporate culture doesn't value balance or having any "bad times" in your life.
...companies still use it ? Thought microsoft taught industry a lesson here
The process itself was flawed because people had a tendency to stay in the rank where they were submitted because frankly managers didn't have the time, energy or interest in really talking through every employee. It was just easier to rubber-stamp the whole process and move on. Also it was nearly impossible to get rid of the process because of course people who had succeeded in it didn't want change. I couldn't in good conscience continue like that, so I took over rankings for my location. It went from one day to a week-long process and I made people vote anonymously and incorporated a ranked choice system and then aggregated the results. It was a far more time consuming process but IMHO fairer. I still hate forced ranking despite succeeding in it. It becomes part of your job - managing to the metric.
Vestiges of it remain because people still don't know how to distribute compensation without ranking people, particularly for employees that have the same review rating.
But yes, all major companies still do something like this AFAIK, just with a larger population size now. Speaking from my experience at Amazon, the target was around 4%, but this was across entire departments. My department was 300 people. Is it possible that you can have a department of 300 people where you can't find 12 people who are below-average performers? Yes. Is it likely? No.
I don't know how many people the lead in the article managed. If it was a team of 15, he has a legitimate gripe. If it was a team of 200, not so much.
Yes, managers are supposed to keep that from happening but this policy tells them that a) you don’t trust them to do their job, b) their job now focuses on protecting their people from the system, and c) being busy playing the same game on their own personal behalf.
I mean, below average performers will be around 40-60% just because how averages work.
The problem is that the question is "is their performance low compared to team", while the question you should be answering is "is their performance low compared to the market.
If team's management is on point with hiring the "worst" won't be that much worse than average anyway, so if you want to replace the bottom few % you essentially waste time as new hires might be at same productivity level.
On other side if you recruiting sucks and 20% of the dept does the 80% of the job, it isn't going to help either.
If your team leads could be trusted to measure appropriately, you wouldn't need stack rankings because they'd correctly identify those under-performing. Stack ranking is conceding that your managers can't properly measure performance. Shouldn't that reflect more on management?
so whatever psychopath games they are playing, its not working.
Dead Comment
So managers everywhere who only saw lots of headlines thought it must be cool just like Scrum and microservices and decided to adopt it.
I didn't know this existed until now. The problem with this (among others) is that I can never personally know where I stand. If I'm doing well, but being excelled by others, I still get penalized. There's no guarantee, it's almost random.
Also, I now have a new question to ask in interviews, and if it does exist in a hiring company, I can negotiate by declining bonus pay performance reviews and in lieu of ask for higher wages (not that any company would go for this).
But if times are good and everyone is performing well enough and contributing, why get rid of them? If they're not performing well enough, then you shouldn't need stack ranking to figure that out, and you don't need to arbitrarily set a percentage of employees to put on performance improvement plans or fire.
Yep, Currently working for an absolute horror show of an employer who proudly uses stack ranking during yearly and half yearly reviews.
Cant wait to move on from this place.
In fact 5% is quite a low estimate in my opinion. I would estimate it 20%, easily 1 in 5
Maybe there is a bigfoot out there but I will refuse to believe until I see one
Stack ranking seems to be a solution for managers that are not doing their job. Either because they are trying to be nice or oblivious or just don't care. Since this guy quited to me it sounds like the first.
Deleted Comment
Oh, really? I haven't followed this kind of stuff in a while. So they're good now?
How would they rank for software development, in general? Game studios are notoriously subpar when you start comparing them to FAANG, big enterprise sotware development shops, etc.
I don't think this has changed substantially. It's supply and demand: lots of people want to build games for the sake of building games. At least in theory it sounds more exciting than maintaining whatever microservice for a FAANG. So, because there's a lot more supply than demand, they get paid less.