Well written and put. I also personally witnessed and experienced all of this, and more, during my time in Amazon operations and HR. Keep shining the light on these dark corners.
Missed one important thing. GM implemented rank and yank because they were secretly on the edge of bankruptcy. They managed to distract from their financial troubles through laying off the bottom 10% every year for longer than they should have been able to manage.
Hey this is a great input. GM is a fascinating case study. The company was stuck with an elevated fixed cost basis, then stopped turning a profit because it wasn't making good or even safe products, then got exposed by the financial crash because it was supporting its production with profit from its financing companies. But underneath it all was that unbreakable basis of fixed costs, including labor costs which were tough to reduce because of union protections.
So to your point, the only way the company could lawfully reduce headcount once it cashiered its bloated temp workforce was to cook up performance-justified sackings.
Many morals to this story, but one is that if you can't afford to pay people a good wage and still make a profit, your business model doesn't work anymore. Oh, and if you don't want to deal with the added inflexibility of union negotiations, or find yourself tunneling under union constraints, maybe just treat your workers better in the first place. Amazon is tempting broad unionization with its current and trending working conditions. I'm not certain the business model can sustain itself in that world, but a lot of corruption will happen along the way in vain attempts to keep it afloat.
What's even more interesting about GM is that it has returned to the practice of Rank and Yank. In its Automation Division, there was a "Minus" rating target of 12.5% in its most recent talent review. Experience tells us those with a Minus will be managed out.
Great stuff. I saw this first-hand when I was at Amazon a decade ago, and previously at Microsoft. What I see at Amazon is they are happily going down the Ballmer iterations of rank-and-yank with equally toxic effects; at one point under Ballmer everyone was put into one of 5 quintiles... bottom quintile = fired or PIP. But what also happened was that everyone in the 2nd and mid quintile found that while technically they could change jobs for greener scenery or more opportunity, nobody would take them because they weren't a top performer. The people in the top two quintiles of course took full advantage of this. Seeing this, what managers did was double-down on their defensive mechanism against the curve - they intentionally over-hired, so that there would be enough disposable fungible people to give bad ratings to so they could give good ratings to those they liked.
Just a Jassy anecdote - early in my career there, so like 15 years ago, I was on a kaizen project with a lot of other sr leaders on the annual review process, sponsored by Jassy and Galbatto (head of HR at the time). Our main accomplishment was to move the time to write the annual review to be done BEFORE managers assigned ratings, vs after which is how it had been scheduled. The actual review doc still had little to do with the rating (which I do think is fair in most cases - a manager should know how their people are performing and doesn't need some self review doc to prove the case; it's just more useful when it's a new manager or some such), but at least it was more palatable. For Jassy, something he talked about with passion was the forced curve - he loved it as he felt it was a mechanism to force people to make choices.
Personally, I took this as tacit admission that manager training and delivery wasn't a thing Amazon would do... so rather than help a new manager really understand how to set expectations, give feedback, measure progress, etc. it was just much easier to have a curve and call it good.
Some choices and philosophies of CEOs are more revealing than others. Jassy's need to force a curve reflects his view of the people who work at Amazon. He clearly does not think highly of his own workforce if he believes they need a forced curve to compel them into a strong work ethic. But this is consistent with his other actions, to include the most recent shareholder letter making no mention of employees at all.
I think it will become impossible for Amazon to maintain a high talent bar if the culture doesn't turn around rapidly. The problem with a decrepit work culture is that the strong leaders jump ship and strong performers stop joining. This leaves too little strength in the organization to reverse the problems, and from there it just degrades sharply.
This is some of what Bezos was referring to when he talked about Day 2. Organizations need to stay disruptive and restless. When they start settling for mediocrity and protecting themselves instead of customers, things get ugly.
There are a lot of things you can get away with when everyones stock options are shooting to the moon every year that you can't get away with when they don't.
It's a complex topic. It's also a popular attack surface.
The underlying idea to focus on both high and poor performers continuously and proactively is solid. But as is common - idea good, execution bad.
There are many poor and unqualified people managers, and as a result the idea gets mechanically applied instead of being a helpful framework. That leads such aberrations as destroying high performing teams or hedge hiring.
What would be more important, is to recognize that people management is a skill and trade akin to all the other functional trades in the organization. Not everyone is good for it, and no one should be forced to do it. Yet, the lack of better career path design and better org designs doesn't leave anyone a choice. People who should never be in charge of people end up in the chair regardless.
One solution would be to separate people and functional management. We're matrixed organizations already, so this wouldn't be difficult. Have people managers focus on people (and with that I don't mean HR partners, they need to remain separate to deal with logistics and risk management), and functional managers who focus on product and delivery.
PS: I've been gone from Amazon for many years now, so my first-hand experience is dated. Back then I didn't experience or sense the gladiator games version of performance reviews, and had positive experiences as an individual and as a people manager. In particular I appreciated that there was a professional attitude about working under high pressure and doing high risk work. You were accountable for performance, but in my years there I only know one of person fired for causing a prolonged outage. You cannot retain top talent doing difficult work if you constantly fear for your job on the slightest mis-step. But from what I have read, today's culture is apparently quite different and much less appealing.
My experience was limited to fulfillment ops, but my observation is that a sharp divergence from the prior culture started around 2.5 years ago as we came out of the pandemic and Jassy took full control. On the subject of talent specifically, '22 was the first year I can recall that we were forced to do a complete stack ranking in every organization. This seemed to come from insistence on hitting an attrition target, and that insistence became more explicit and obvious in the next reviews.
I agree the underlying impulse is healthy. I've just never believed that in a context of continuous bar-raising, we must always assume a slice of dismissal-worthy performance. Nothing worse than actually doing the dirty work *before* talent review to manage someone out, only to have to stick someone else in a bucket to hit a quota.
All that said, I would feel different and take a different position if Amazon would simply be honest. Explain why there is a commitment to attrition and stand by it.
We won't see that of course, because it is a basically indefensible ethical perimeter. Honesty would expose the absence of true necessity, leaving only coercion as a motive.
Not entirely surprised. I remember from all the s-team meetings I attended Jassy came across as his own island in the room. That was the early days of AWS, and it was less connected to the rest of the company, as none of the core infrastructure ran on AWS at that point. But it was also in the attitude. Jeff could have a position out of left field that you didn't expect, but it was always based on some core beliefs and once you understood the reasoning, you were on board and appreciated the perspective. I can remember two distinct moments like that. With Jassy, while I never worked with him directly, I get the impression that he has much fewer deep personal principles, but just likes to follow the herd instinct of many of the problematic directions. RTO being a prime example. He says it's all grounded in data, but you can always find data for your story. With Jeff it was an innate intuition of the data, and we usually saw it validated in weblabs later, even if nobody thought it could possible be right. The two of them are not cut of the same wood. That's a shame for Amazon.
Wilke, which I had much more regular interactions with, was a lot more like Jeff in many regards, and would likely have been more in line with the old Amazon. In 10-15 years we may look at the succession choice and call it non-optimal. Of course I'm not privy to the discussions or what factors the board was optimizing for at the time.
Ironically this is how the U.S Military does it. The difference is that it is explicit. As statutes essentially say how many people in each paygrade you can promote, and how long you can stay in a particular paygrade. rank and yank is basically up or out. How many times did I see an great E-4 or a great O-3 whom were really good at their jobs but had no desire to go up a rank be given walking papers after a certain period of time.
I was going to ask if anyone saw similarities to the Military rank system and "forced distribution".
The counter is the Rank system used to inappropriately "save" the career of an underperformer vs validate and recognize the performance of a superstar.... Well we need to rank them high cus they are coming into zone or whatever justification, your guy will get same treatment when it's their turn.
Absolutely, excellent point on the O3 or E3 good at their job - technical expert but no desire to manage troops, but that is viewed as dropping the pack or giving up.
Well written and put. I also personally witnessed and experienced all of this, and more, during my time in Amazon operations and HR. Keep shining the light on these dark corners.
Thanks Greg, and thank you for reading.
Missed one important thing. GM implemented rank and yank because they were secretly on the edge of bankruptcy. They managed to distract from their financial troubles through laying off the bottom 10% every year for longer than they should have been able to manage.
Hey this is a great input. GM is a fascinating case study. The company was stuck with an elevated fixed cost basis, then stopped turning a profit because it wasn't making good or even safe products, then got exposed by the financial crash because it was supporting its production with profit from its financing companies. But underneath it all was that unbreakable basis of fixed costs, including labor costs which were tough to reduce because of union protections.
So to your point, the only way the company could lawfully reduce headcount once it cashiered its bloated temp workforce was to cook up performance-justified sackings.
Many morals to this story, but one is that if you can't afford to pay people a good wage and still make a profit, your business model doesn't work anymore. Oh, and if you don't want to deal with the added inflexibility of union negotiations, or find yourself tunneling under union constraints, maybe just treat your workers better in the first place. Amazon is tempting broad unionization with its current and trending working conditions. I'm not certain the business model can sustain itself in that world, but a lot of corruption will happen along the way in vain attempts to keep it afloat.
What's even more interesting about GM is that it has returned to the practice of Rank and Yank. In its Automation Division, there was a "Minus" rating target of 12.5% in its most recent talent review. Experience tells us those with a Minus will be managed out.
Great stuff. I saw this first-hand when I was at Amazon a decade ago, and previously at Microsoft. What I see at Amazon is they are happily going down the Ballmer iterations of rank-and-yank with equally toxic effects; at one point under Ballmer everyone was put into one of 5 quintiles... bottom quintile = fired or PIP. But what also happened was that everyone in the 2nd and mid quintile found that while technically they could change jobs for greener scenery or more opportunity, nobody would take them because they weren't a top performer. The people in the top two quintiles of course took full advantage of this. Seeing this, what managers did was double-down on their defensive mechanism against the curve - they intentionally over-hired, so that there would be enough disposable fungible people to give bad ratings to so they could give good ratings to those they liked.
Just a Jassy anecdote - early in my career there, so like 15 years ago, I was on a kaizen project with a lot of other sr leaders on the annual review process, sponsored by Jassy and Galbatto (head of HR at the time). Our main accomplishment was to move the time to write the annual review to be done BEFORE managers assigned ratings, vs after which is how it had been scheduled. The actual review doc still had little to do with the rating (which I do think is fair in most cases - a manager should know how their people are performing and doesn't need some self review doc to prove the case; it's just more useful when it's a new manager or some such), but at least it was more palatable. For Jassy, something he talked about with passion was the forced curve - he loved it as he felt it was a mechanism to force people to make choices.
Personally, I took this as tacit admission that manager training and delivery wasn't a thing Amazon would do... so rather than help a new manager really understand how to set expectations, give feedback, measure progress, etc. it was just much easier to have a curve and call it good.
Some choices and philosophies of CEOs are more revealing than others. Jassy's need to force a curve reflects his view of the people who work at Amazon. He clearly does not think highly of his own workforce if he believes they need a forced curve to compel them into a strong work ethic. But this is consistent with his other actions, to include the most recent shareholder letter making no mention of employees at all.
I turned down a job offer from Amazon because of things like this.
I think it will become impossible for Amazon to maintain a high talent bar if the culture doesn't turn around rapidly. The problem with a decrepit work culture is that the strong leaders jump ship and strong performers stop joining. This leaves too little strength in the organization to reverse the problems, and from there it just degrades sharply.
This is some of what Bezos was referring to when he talked about Day 2. Organizations need to stay disruptive and restless. When they start settling for mediocrity and protecting themselves instead of customers, things get ugly.
There are a lot of things you can get away with when everyones stock options are shooting to the moon every year that you can't get away with when they don't.
It's a complex topic. It's also a popular attack surface.
The underlying idea to focus on both high and poor performers continuously and proactively is solid. But as is common - idea good, execution bad.
There are many poor and unqualified people managers, and as a result the idea gets mechanically applied instead of being a helpful framework. That leads such aberrations as destroying high performing teams or hedge hiring.
What would be more important, is to recognize that people management is a skill and trade akin to all the other functional trades in the organization. Not everyone is good for it, and no one should be forced to do it. Yet, the lack of better career path design and better org designs doesn't leave anyone a choice. People who should never be in charge of people end up in the chair regardless.
One solution would be to separate people and functional management. We're matrixed organizations already, so this wouldn't be difficult. Have people managers focus on people (and with that I don't mean HR partners, they need to remain separate to deal with logistics and risk management), and functional managers who focus on product and delivery.
PS: I've been gone from Amazon for many years now, so my first-hand experience is dated. Back then I didn't experience or sense the gladiator games version of performance reviews, and had positive experiences as an individual and as a people manager. In particular I appreciated that there was a professional attitude about working under high pressure and doing high risk work. You were accountable for performance, but in my years there I only know one of person fired for causing a prolonged outage. You cannot retain top talent doing difficult work if you constantly fear for your job on the slightest mis-step. But from what I have read, today's culture is apparently quite different and much less appealing.
My experience was limited to fulfillment ops, but my observation is that a sharp divergence from the prior culture started around 2.5 years ago as we came out of the pandemic and Jassy took full control. On the subject of talent specifically, '22 was the first year I can recall that we were forced to do a complete stack ranking in every organization. This seemed to come from insistence on hitting an attrition target, and that insistence became more explicit and obvious in the next reviews.
I agree the underlying impulse is healthy. I've just never believed that in a context of continuous bar-raising, we must always assume a slice of dismissal-worthy performance. Nothing worse than actually doing the dirty work *before* talent review to manage someone out, only to have to stick someone else in a bucket to hit a quota.
All that said, I would feel different and take a different position if Amazon would simply be honest. Explain why there is a commitment to attrition and stand by it.
We won't see that of course, because it is a basically indefensible ethical perimeter. Honesty would expose the absence of true necessity, leaving only coercion as a motive.
Not entirely surprised. I remember from all the s-team meetings I attended Jassy came across as his own island in the room. That was the early days of AWS, and it was less connected to the rest of the company, as none of the core infrastructure ran on AWS at that point. But it was also in the attitude. Jeff could have a position out of left field that you didn't expect, but it was always based on some core beliefs and once you understood the reasoning, you were on board and appreciated the perspective. I can remember two distinct moments like that. With Jassy, while I never worked with him directly, I get the impression that he has much fewer deep personal principles, but just likes to follow the herd instinct of many of the problematic directions. RTO being a prime example. He says it's all grounded in data, but you can always find data for your story. With Jeff it was an innate intuition of the data, and we usually saw it validated in weblabs later, even if nobody thought it could possible be right. The two of them are not cut of the same wood. That's a shame for Amazon.
Unfortunately I think the founding vision and spirit departed with the founder.
If I were being charitable, I might say we're still in a flux between two steady states. But I've seen too much to be charitable at this point.
Your observations about the S-team interactions are fascinating, and they align with what I have heard from many others.
Wilke, which I had much more regular interactions with, was a lot more like Jeff in many regards, and would likely have been more in line with the old Amazon. In 10-15 years we may look at the succession choice and call it non-optimal. Of course I'm not privy to the discussions or what factors the board was optimizing for at the time.
Ironically this is how the U.S Military does it. The difference is that it is explicit. As statutes essentially say how many people in each paygrade you can promote, and how long you can stay in a particular paygrade. rank and yank is basically up or out. How many times did I see an great E-4 or a great O-3 whom were really good at their jobs but had no desire to go up a rank be given walking papers after a certain period of time.
I was going to ask if anyone saw similarities to the Military rank system and "forced distribution".
The counter is the Rank system used to inappropriately "save" the career of an underperformer vs validate and recognize the performance of a superstar.... Well we need to rank them high cus they are coming into zone or whatever justification, your guy will get same treatment when it's their turn.
Absolutely, excellent point on the O3 or E3 good at their job - technical expert but no desire to manage troops, but that is viewed as dropping the pack or giving up.