It’s Not Surveillance, It’s Performance Improvement
Chapter 11 in my ongoing story about workplace surveillance
< Previous Chapter | First Chapter | Plot Summary | Next Chapter >
It only took six months for Stamper and I to catch up again, and that put us in prime hiking season, so we agreed to meet up at the Rachel Lake trailhead to try a new-to-me summit, Alta Mountain. I offered to meet her at a park-and-ride and share the ride, but she said she’d prefer to drive alone.
“Remember, you’re still under NDA,” was the first thing she said to me as I pulled in next to her at the trailhead and we both hopped out of our cars.
“Well, good to see you too!” I laughed. “I don’t think I spilled any beans after our first conversation.”
“None that I’ve heard about. But then I hadn’t told you anything particularly interesting,” she teased.
“Ooh, does that mean I get the juicy stuff today? This should be fun.”
“Fun,” Stamper sighed, suddenly serious. “That’s one way to describe it.”
“Oh sorry, I didn’t mean to hit a sore spot,” I said. “And listen, I’m sorry we got a little heated last time. I promise, I’m not going to be such a pain in the ass this time.”
“You were fine,” she sighed. “You were just hitting a little too close to home.”
It was a long hike we were setting out on: a couple miles relatively flat through the woods, then up a set of switchbacks to reach Rachel Lake. Above Rachel Lake we followed a long and glorious ridge to Alta Mountain. Seven miles one way. The long valley-to-summit hikes were my jam—and this one afforded plenty of time for one of the most interesting conversations I’ve had in a long time.
Stamper put an abrupt end to the light-hearted chit-chat portion of the hike barely a mile in.
“Remember your Stasi comment from our last hike?” she said.
“Yeah, I do, and I’m sorry—that was just me being jealous.”
“More like prescient!”
“What? Seriously?” I asked.
“Seriously … oh, well, semi-seriously. I’m just a little worried about where we’re going. Can we talk about it? Like, can you help me think about what’s going on?”
“You bet Kate,” I said, softening the conversation with her first name, which I rarely used. “And don’t worry, I know I’m under NDA.”
She smiled and accepted my olive branch.
“So, not long after our last hike, we got a new director in Human Dynamics,” she said.
“Replacing that guy who poached you?”
“Ha ha, yes, you remember! Anyway, he’s a long-time Amazon guy—like one of the first 200 employees, and he’s been in retail forever. Legend has it that he was one of the people who first unlocked the data tracking that make Amazon product recommendations so precise.”
“I bet that sounded great to you, having someone with that kind of experience.”
“Yeah, it did—I mean, that’s what I’ve been hoping to do for like years. I want to get people information and training just when they need it, but only when they need it. But I need some smart data people and coders to get there. Story of my life!” Stamper said.
“So does he bring that?”
“Oh yeah, in spades. He’s all about using data to understand behavior and generate actions, and he brought along a couple top devs.”
“Isn’t he slumming to come over to Human Dynamics?” I asked. “I mean, those business-data guys usually find training and behavior change kind of squishy.”
“That’s what I thought too, but he said that once the S-team—that’s the upper level leadership at Amazon—learned about what we were doing with cybersecurity behavior, they all saw the opportunity to do something that would help the company and he jumped at the chance to come over here.”
“God, that’s cool,” I said. “Does that mean you’re getting even more resources to do what you want?”
“Well first, yes, I mean resources are not the issue! Holy shit, it’s like whatever we want to do, we can do. But I’m not so sure on the cool part.”
“I don’t know, getting to do whatever you want sounds pretty good to me!” I said, remembering all the things I had shut down over the years because I knew they would cost too much.
“It does, right up until you realize the guy you’re working for is a sociopath!”
“Oh jeez, what are you talking about?”
“Let me tell you about Mitch Cascade,” she sighed.
“What, you’re working for a Disney character?”
“I know, I know, it’s weird, but it’s the perfect name for the guy,” she said.
Mitch Cascade walked into Stamper’s office like he owned the world. “Hey,” he said as he knocked on the open door and waltzed in, “you must be Kate Stamper.” He grinned broadly and stuck out his hand. “I’m your new partner in crime.”
Stamper was immediately struck by how polished he was. He stood about 5’10”, his beautifully graying hair trimmed close, but not too close to hide the slightest wave. He wore tailored blue jeans, no label, and a white button down shirt under a navy blue cashmere sport coat; his shoes were low-cut “tennis shoes,” only made of a lovely mahogany leather. It didn’t take too many days for Stamper to realize that this was his uniform, and it varied only in the shade of the shoes and the color of the shirt.
Mitch Cascade had a way of looking you right in the eye as if he was waiting for you to say something fascinating. As long as he thought you were on his side, his grin was perpetual. Cascade could have charmed the skin off a snake—and he never stopped trying.
“Oh hi,” said Stamper. “I wasn’t expecting you until next week.”
“I know, I know,” said Cascade, “but I was dying to get started. I hear you’re doing really cool stuff here. Are you ready to do more? Are you ready to blow this thing up?” He beamed at her. His teeth were perfect.
“That’s how it started,” she told me. “He came in like his sole purpose in life was to help me extend what I was doing.” They spent their first hour together with him just listening to Stamper describe how they were gathering data, how they were creating content to “fix” the human errors that the data showed, and the ingenious ways they were nudging employees to be better at cybersecurity and privacy.
“Kate, this is great,” he said, “just great, but it’s just the start. I want to help you take it further. I mean, have you thought of all the other ways that we could help our employees be better Amazonians?”
“Oh definitely,” Stamper interjected, “like better communicators, better at follow-up …”
“And better managers,” Cascade joined right in, “better innovators, better at giving feedback and direction …”
“And we can just help people not make mistakes,” said Stamper. “Like why would there ever be a mistake on expensing or reporting, if we can just interject feedback into the experience?”
At this point, they both just leaned back and laughed. They’d barely been talking twenty minutes when they were completing each other’s sentences.
“For a time it was just like that,” Stamper told me. “The first couple months were great: with his encouragement and a lot of times his direct input, we’d take up an area of human performance and then identify what tools we already had in place to identify leading indicators, understand the baseline behaviors, then develop content, and then test all the different ways we had to nudge human performance in the right direction. Our capacity to run tests was just unreal—I mean, with over a million employees, we could do sample runs on our BMTs …”
“Woah woah woah Stamper,” I said, “remember: I don’t know the acronyms.”
“Oh, right,” Stamper laughed. “BMTs: Behavioral Modification Taps. We called them ‘taps’ because we wanted them to just feel like little taps on the shoulder—just a brief, barely noticeable reminder to modify your behavior in the right direction.”
She described how they’d develop several different styles of “taps” that they’d try out on people, usually groups of 1,000 at a time, to determine “how hard they had to push” to get people going in the right direction. Within a week to 10 days, they had a pretty good sense of what worked and what didn’t, and they’d just put that stuff that worked right into action and move on to the next behavioral target. They always figured these were “two-way doors,” to use a favorite Amazon metaphor: there was nothing they were doing that couldn’t be reversed if it didn’t work.
“Dan, I never met an executive who was willing to move so fast, to just put stuff into action,” she said. “Remember how slow everything went at Wizards, how scared everyone was to take big steps forward? This was the opposite of that. It was pretty thrilling. It’s like Cascade must have had a lot of backing from the top, because there was just no hesitation. It was just, do it. Sometimes it went so fast it made my head spin.”
Soon enough, they started to generate dashboards on how they were doing at modifying employee behaviors. Stamper described one of them to me:
“We decided that we wanted to make real improvements in discriminatory language,” she explained it. “Like, we wanted to end sexist and racist language in employee communication. Great, right? So we set up flags on all the channels of communication—email, Slack, Chime meetings, even phone calls. So, if people said or wrote anything discriminatory, we could step in.”
“What do you mean, ‘step in?,’” I asked.
“Okay, say you send an email that says, ‘The ladies in HR are slowing things down,’ or some sexist bullshit like that. Well, you go to send that email, and we block it—like, keep it from sending—and launch a pop-up message that says, ‘Remember, it is Amazon’s policy not to use language that diminishes others based on gender, race, ethnicity, or sexual orientation.’ And so then you get to change that language and you don’t offend anyone. Pretty cool! So we implement this and we see that it dramatically reduces the use of discriminatory language in internal communication.”
“So basically you’ve figured out how to keep people from being assholes!” I joked.
“Yeah, that’s about it,” Stamper laughed.
“And you do this on calls and web meetings too?”
“Yeah, it’s not quite as direct there, but we can track what is said in real time, and if there’s something bad, well, it’s not like we can stop it from being said, but we can immediately send a message to their phone or in their chat that identifies the error … and then we can follow up later, and also track whether they continue to make these errors or they improve.”
“Stamper, this sounds pretty fucked up,” I said. “I mean, you’re listening to people’s calls and immediately pinging them to tell them to stop?”
“You know, I get what you’re saying, but I think we’re okay here: there’s been this big push in the company to let everybody know that discriminatory language isn’t okay, so what we’re really doing is just helping make people aware when they are stepping over the line. It’s not like we’re really penalizing people, we’re just gently nudging them to do the right thing.”
“You said you’re not ‘really penalizing,’” I observed. “Does that mean you’re sort of penalizing?”
“Well, I guess, yeah, but we don’t think of it as penalizing. Basically, we aggregate all the data on each individual and when we see a systematic pattern, we have some escalations in place to correct them.”
“Escalations?”
“You know, like gradually strengthening ‘taps.’ We call them TEs—tap escalations. Most people never go beyond a Level 3 TE.”
“Oh god! What happens at a Level 3 TE?” I asked.
“Oh, at Level 3 for a discriminatory tap, we end the call or meeting where the infraction occurred, or shut down their email, and they are called into an immediate meeting with an HD specialist. It’s pretty effective. We’ve never had to take the next step.”
“Which is …?”
“Termination.”
“Wow! That’s intense.”
“I know, but honestly, kind of cool, right? I mean, we’ve figured out how to put an end to discriminatory language in the workplace. In less than six months! Can you believe it?”
“Well, yeah, I can believe it!” I said. “It’s kind of where I feared this stuff was going all along. It sounds to me like you’re escalating the hell out of employee surveillance and you’re totally fine with it. What the hell?”
“It’s not employee surveillance Dan! It’s performance improvement.”
“I know that’s what you’re calling it, but that’s not what it looks like.”
“Look, I have no problem defending this part …,” she trailed off.
“But?” I asked.
“But some of my team is starting to feel a little weird about it.”
In your research I would be curious to understand the relationship between surveillance and the size of the company, as well as its relationship to Silicon Valley. Either way, I'm hooked into this story thread now, and I'm excited to read more. Great job keeping up the suspense, Tom!
I’m glad you’re hooked--that’s a good sign for me! Once it’s all done, I’ll do a piece on what’s real and what’s not, kind of summarizing everything that I’ve read.