Don't be Creepy. Employee Surveillance Backfires.
Just because employee spyware exists, doesn’t mean you should use it.
This week, Adam Satariano published an article called “How My Boss Monitors Me While I Work From Home.” He describes a 3-week experiment with HubStaff, an employee surveillance tool. The takeaway: “ick.”
Coincidentally, on the same day, Github announced Github Insights, for tracking engineering team performance. Despite consternation on Hacker News, this is a much saner way to get a pulse on team performance.
Both tools attempt to accomplish the same goal (“team performance visibility,” especially for management). And both are a little creepy. But, the former fails in ways the latter handles quite well.
So what’s the difference?
First, consider 3 principles:
- If you track something, you’ll see it improve. It works for team OKRs. It works for Fitbits. This is well-documented and generally a good thing.
- “Getting what you ask for” !== “getting what you want.” This is a fundamental challenge with any metric. There’s always distance between the measure and the underlying phenomenon. So, just because a metric improves, doesn’t mean underlying behavior has changed for the better.
- Nothing good comes from treating people like robots. Grossly violating employee autonomy is a quick way to turn them against your cause.
How does this play out with Screenshot Tracking?
Satariano’s describes Hubstaff:
Last month, I downloaded employee-monitoring software made by Hubstaff, an Indianapolis company. Every few minutes, it snapped a screenshot of the websites I browsed, the documents I was writing and the social media sites I visited. From my phone, it mapped where I went, including a two-hour bike ride that I took around Battersea Park with my kids in the middle of one workday. (Whoops.)
One main feature of Hubstaff is an activity monitor that gives managers a snapshot of what an employee is doing. Broken down in 10-minute increments, the system tallies what percentage of time the worker has been typing or moving the computer mouse. That percentage acts as a productivity score.
It’s true that if you evaluate people based on “10-minute mouse-movements” and “productive looking screenshots”, you will see more mice moving and productive looking screenshots.
What won’t you see? Employees doing more work (or better work).
Even if “it’s a jerk move” doesn’t dissuade you from creeping, be aware that it’s also ineffective.
How does Screenshot Tracking sabotage productivity?
- Screenshot or activity tracking implies that both of these behaviors will be rewarded: a) moving the mouse around while looking at “serious” things, or b) doing actual work. Because “moving the mouse” is easier than “doing actual work”, expect to see more “moving the mouse.”
- To the extent that “moving the mouse” conflicts with “doing actual work,” (e.g. reporters get less credit for untracked behavior like calling sources), expect to see more “moving the mouse.”
- And, finally, now that employees are pissed off by the now-apparent gaping void of trust, they have even less of an incentive to optimize their performance in good faith, so expect to see more “moving the mouse.”
The worst part: The manager will indeed see more “moving the mouse.” So, in addition to having employees who may-or-may-not be fully committed, the manager now has an even more opaque view of whatever issues are actually causing the team to slip.
Screenshot Tracking on Upwork:
One engineering lead I spoke with had previously hired programmers on UpWork, a freelance site that offers the option to screen-track workers.
He said: Screen-tracking the freelancers: “almost uniformly led to worse work”
“You have this false sense that you’re measuring their output. It lulls you into a sense of complacency. Plus there’s all these extra screenshots you’re expected to look at. If you can’t measure results by what this person is producing, there’s something deeply wrong.”
He also noted: “But my boss loved it.”
It’s not just Screenshots…
Another engineering leader reflected on an early-career experience. He had an overzealous former-manager who insisted on strict time-tracking.
“It was a nightmare. The manager set expectations of 35 hours per week at the keyboard. For every bit of work you do, you track the task, you track the amount of time you spent on it.
He was looking at those numbers and he would bring them up often. So what I found myself doing - this was early in my career - I just fucking lied a lot.
It created stress. It created a bad habit where I had to hit a target number and it wasn’t about delivering code quality. It wasn’t about delivering value. So I just gamed the system.”
It’s not just tracking programmers…
Harvard Business School’s Ethan Bernstein has studied workplace surveillance extensively. In one study (“The Transparency Paradox”), he embeds a team of grad students on a factory floor, and they document even more surprising ways that close supervision can undermine performance:
First the [embedded researchers] were quietly shown ”better ways” of accomplishing tasks by their peers - a ”ton of little tricks” that ”kept production going” or enabled ”faster, easier, and / or safer production.”
Then they were told, ”Whenever the [customers / managers / leaders] come around, don’t do that, because they’ll get mad.” Instead, when under observation, embeds were trained in the art of appearing to perform the task the way it was ”meant” to be done according to the codified process rules posted for each task.
Because many of these performances were not as productive as the ”little tricks,” I observed line performance actually dropping when lines were actively supervised.
Again, it doesn’t pay to creep!
What’s different about GitHub Insights?
It plays nicely with the principles.
- If you track something, it’ll improve.
- “Getting what you ask for” !== “getting what you want.”
- Don’t treat people like robots.
GitHub Insights analyzes engineering teams’ shared code repository to approximate team behaviors that (at least loosely) correlate to team productivity. Because Git repos are essentially big, living records of a teams’ interactions with their codebase, they can be a telling source of data on team behaviors. Other similar “Engineering Intelligence” tools include CodeClimate Velocity, Waydev and SourceLevel.
Unlike invasive methods like Screenshot Tracking, Git-based analytics give teams a way to measure performance that 1) focuses on metrics with a stronger relationship to desired behaviors, 2) is much less creepy and 3) focuses on teams, not individuals.
The metrics being emphasized are more representative of actual productive behaviors worth improving. It’s certainly still possible to game metrics like “code review speed”, but much harder than “moving your mouse.”
It’s much less creepy. Everything in Git is already accessible to the team. If you were so determined, you could calculate the same stats by hand.
The focus of the measurement is the team, not an individual. It’s much easier to buy-into. It’s ego preserving. No one is getting called out individually. And, because this approach preserves trust, people are more likely to optimize behaviors against the target metrics in good faith.
Tracking and measuring are fine. It works. But don’t creep. Favor me that reflect meaningful behaviors. Respect employee autonomy. Focus on improving team performance, not calling out individuals. Be nice. Be human.