Ethan Evans × Ryan Peterman · Amazon VP: Stack Ranking & PIPs, Working With Bezos, His Promotions
The dot-com bubble had burst and 9/11 had just happened when Evans found himself out of work. It was the second time he had left a startup. On paper it was a layoff, but he was the only person laid off [00:07:50]; he calls it being fired. At an interview he walked through his accomplishments, and the interviewer said it all sounded fine, but he struggled to believe that if Evans were that valuable, those companies wouldn't have found a way to keep him [00:12:32].
After it happened twice, I damn sure looked at like, who's the common element in this? It's me.— Ethan Evans · [00:07:50]
That is the key to the whole episode. Everything he says later about promotions, PIPs and HR comes from someone who was burned by the story told about him and later sat in the chair that tells the story.
Evans grew up in Ohio and, as a kid, mailed Apple a physical letter asking how to work there; the answer was that Apple hired engineers from only five universities. He went to Carnegie Mellon, then moved through startups in high-speed networking, search and audio streaming, and Amazon hired him on the strength of the music-streaming work to build Prime Video. By his own public bio he spent about fifteen years at Amazon, ran the 800-person Appstore, retired as a VP in 2020, and now teaches people how to clear the promotion hurdle. At the end he says plainly that if this straight-talk style works for you, that is what he sells [00:52:34]. Keep that in view: "I'll say what others won't" is itself his brand.
Ryan Peterman was an engineer at Meta (Instagram) who also climbed quickly, and his show keeps coming back to three questions: what each level demands, how you get promoted, and what you regret. When the episode went up, the Amazon Appstore that Evans mentions had already announced in February that its Android version would close on August 20, so his "I saw that coming ten years ago" landed right on schedule.
Young Evans's problem was concrete. His very first performance review said his solution to problems was sometimes to knock people's heads together [00:05:16]. Both exits came from the same kind of conflict. A product lead said the plan couldn't be built in time; Evans thought he wasn't committed, but he was right. A sales VP was selling features that didn't exist yet; Evans fought him, missing that without that contract the company would go under [00:05:56].
He draws two mechanisms from this, both sharper than "have more EQ." First, push too hard and the people who disagree don't keep arguing; they stop talking to you and start talking to others about you. Second, he hadn't built relationships with B, C and D, so when A went to them, nobody came back to tell him; they just nodded that yes, he's a problem [00:10:45]. Engineers think being right is enough; in practice people will pick a worse solution that feels more comfortable.
He isn't arguing for being a pushover. He likes a phrase someone taught him, "strategically annoying" [00:09:51]: have an opinion and fight for it; what matters is how you fight. His best demonstration is what he told his boss when he wanted the director promotion:
My career is very important to me, and I need to know if it's as important to Amazon, because if it's not as important to Amazon as it is to me, I have to think about that.— Ethan Evans · [00:18:37]
It means "promote me or I'm leaving," yet no word of it can be held up as a threat. It worked because his VP was about to leave, and if both of them went the project would fall apart. He then switches to the manager's side for the episode's first "people won't like this": with two roughly equal people, one agitating and threatening to quit and one patiently waiting, he often promoted the agitator, because he could only push one promotion through per quarter or year, and promoting the quiet one meant losing the loud one.
I'm actually willing to tell you the truth and tell you that almost everyone's doing what I'm doing. They're just not telling you.— Ethan Evans · [00:20:09]
Peterman doesn't ask how he knows about "almost everyone," so the line rests on his own observation. His later choice pushes the same skill to its limit: he voluntarily gave up the 800-person Appstore to become the integration liaison for Twitch. Amazon paid about $970 million for it and left CEO Emmett Shear in charge, a man in his early thirties, fifteen or sixteen years younger than Evans. He had no team and could give Shear no orders [00:29:35]. That was the point: practising influence without authority.
Evans splits promotions into "lifetime achievement awards," where the process eventually grinds you up, and outsized events. His director promotion was the second kind. He told his SVP that unless he was ordered not to, he would do the project; the reply was "Great. Hang yourself." [00:17:35] The project was a partnership with TiVo, the DVR company. It worked, and the key line in his promotion document was that without Ethan there would have been no TiVo. Peterman draws the lesson about credit: you don't need to hoard it, you need the narrative "take him out and this doesn't happen" to be obvious.
VP took two and a half years. He calls the method the "magic loop": find out what your manager needs, do all of it, and in exchange your manager does one thing, which is make sure you're rewarded [00:23:43]. He also credits luck: his VP stayed in the role for two and a half to three years and was reorganized away within six months of the promotion. Six months later and he would have started over.
The most useful counterexample is his own failure. Early at Amazon he made a deal with a star new-grad engineer: help me ship our first version and I'll get you from SDE1 to SDE2. Used to startups, where you just told your boss, he didn't know Amazon ran promotions twice a year with deadlines. The launch was a mess and the cycle passed [00:26:56]. The engineer left for another big tech company, and it cost him a year. Evans reaches for Scrum's chickens and pigs: in ham and eggs, the chicken is involved and the pig is committed; the manager is the chicken, you are the pig. So learn the process yourself. Ask your manager whether the document will be done by September 15; if they ask why September 15 matters, that's a warning sign.
Evans's account of Amazon's performance system starts with the goal: a yearly unregretted-attrition target, usually around 7% and as low as 4% in some years, as he tells it. Managers put people on a list, and each one either improves, quits or is fired [00:41:29]. His argument rests on a contrast: the year the goal was removed, managers stopped talking to weak performers. The hardest case is often the mediocre but well-liked person who brings donuts and runs the parties; move against them and the whole team asks why you're trying to get rid of them.
The flip side of a goal is a quota. Once there's a list, HR and your boss keep asking when the person will be gone, and taking someone off a PIP means finding a replacement [00:44:25]. He names the machine's real lever: the first move. Software performance has no objective readout, and the same numbers support opposite stories:
Nothing about software engineering performance is so objective.— Ethan Evans · [00:48:03]
His conclusion: don't expect HR to investigate and rescue you. Fighting a vindictive boss is bringing a knife to a gunfight; either find a way to get along or find a different manager [00:49:30].
The same game is played with business goals. Just before leaving the Appstore he told a peer that this year he would hit every goal and it would be meaningless [00:36:53]. The goal that mattered was competing on Android against Google's pre-installed store; he had no idea how, so he fought not to be measured on it. Bezos's vision for Prime Gaming was to be as big as Tencent; they hit a pile of interim goals without ever being on that path. He was rated well every year.
For my review and my pay, I want goals I know I can nail. And for the good of the company, I actually want goals that I have no idea how to do.— Ethan Evans · [00:40:26]
Put side by side, the two are one structure: the numbers used to judge people and the numbers used to judge projects both sit some distance from what they are supposed to represent, and whoever controls the narrative operates in that gap. That is why big companies carry so many "zombie products": nobody wants to kill them, since that's an admission of failure, and nobody knows how to make them big.
The Bezos-versus-Jassy comparison in the middle lands on the same line. He had roughly 50 meetings with each. When CFO Tom warned Bezos that there was only so much money in the bank, Bezos asked how much that was, because he might want to spend it [00:34:07]. A founder can take the company to Vegas; a hired CEO answers to the board.
Amazon sets a yearly unregretted-attrition goal, usually around 7% and as low as 4%; about ten years ago it was dropped for a year, managers stopped dealing with weak performers, and it was put back. [00:41:29]
Supported
Internal documents seen by the Seattle Times in 2021 show Amazon aims to have 6% of its office workforce leave as unregretted attrition each year, inside his range. The dropped-for-a-year part rests on one anonymous AWS director, with no year given. Amazon denies it has ever stack-ranked.
He has coached about six Amazon employees on PIPs and all were fired; people he didn't coach were fired too; "very few people ever dig out." [00:44:25]
Insufficient evidence
No outcome data on Amazon PIPs is public. The same internal documents give an expectation: more than a third of employees on PIPs will fail. That is a floor and doesn't contradict "nearly all," but his sample is about six people who sought out a coach, which is heavily skewed.
While he was there, Amazon grew from 10,000 people to a million, 100-fold, and revenue grew about 80 times. [00:51:00]
Supported, but overstated
Amazon's 10-Ks: roughly 9,000–12,000 employees at the end of 2004–05, about 1.3 million at the end of 2020, so 108 to 144 times and he undersold it. Revenue grew 45 to 56 times, not 80. And over 400,000 of the people hired in 2020 went into the fulfillment network.