Burnout Tracking Without the Big Brother

Photo: Yan Krukau / Pexels

It usually starts with good intentions. A director who genuinely cares about the team gets worried about burnout, goes shopping for tools, and the market is happy to oblige: keystroke loggers, mouse trackers, software that screenshots people's screens every few minutes. It feels proactive. I'd call it one of the most expensive mistakes a manager can make.

Monitoring software doesn't measure health. It measures motion. It can tell you a keyboard was busy; it can't tell you whether the person behind it is coping or quietly going under. Worse, it tells your team something you never meant to say: we don't trust you. The people most likely to leave over that have somewhere else to go, exactly the people you can least afford to lose. Trust is cheap to spend and painfully slow to earn back.

The line that matters

Two different things get lumped together here. One is watching an individual minute by minute: what they typed, when the mouse moved, how long a window sat idle. The other is asking a team, on a cadence, how work actually feels, then reading the answers in aggregate. The first is surveillance. The second is paying attention, and it never touches anyone's private activity.

The team level is also where the signal lives: burnout rarely announces itself in a 1:1. It shows up as slipping delivery, quieter standups and rising absence, easy to miss until it's a crisis. A recurring, aggregated pulse turns that drift into a trend you can watch instead of a feeling you can't act on.

Ask the team, not the keylogger

The alternative is old and pleasantly boring: the "squad health check" agile teams popularized in the mid-2010s. A short survey, a few named dimensions, repeated on a cadence. Everyone rates things like Workload, Energy and Clarity on a simple scale. No single answer means much; the trend across runs means a lot.

Northlake runs an anonymous Pulse 5 check for its Platform team every few weeks. One run shows Workload falling hard against the last. SquadBear flags the drop, and Marta opens an improvement action with an owner and a due date: rebalance Platform's on-call rotation before the next sprint. Nobody knows who rated what, and Marta doesn't need to. The score just told her where to have the conversation she'd have wanted to have anyway.

Where this lives in SquadBear

Health checks sit under Team Health. Four built-in models cover most teams: Team Health 11 (the classic traffic-light check), Pulse 5 (five quick dimensions: Workload, Energy, Clarity, Support, Satisfaction), Engineering Health 8 for code quality and on-call load, and eNPS. Admins can build custom models at Team Health → Health settings.

Every run picks an anonymity mode: Named, Alias or Anonymous. Whichever you pick, nobody, including admins, ever sees another person's individual rating. And closing an Anonymous or Alias run permanently destroys the link between people and their answers. Not hidden behind a permission. Destroyed.

Trends live at Team Health → Trends, visible to managers and admins: one row per dimension, one column per closed run. When a dimension drops by at least 15% of its scale range against the previous run (0.6 points on Pulse 5's 1–5 scale), every facilitator gets a "score dropped" notification. A rise never alerts anyone.

SquadBear is agent-native, so the whole loop runs from chat too. Two prompts to steal:

"Start a Pulse 5 health check for the Platform team, anonymous, and open it now."

"Pull Platform's health trends and tell me if Workload just triggered a drop alert."

Privacy as a hiring advantage

Senior engineers increasingly ask in interviews how they'll be monitored. They've heard the horror stories. "We don't surveil people, and here's how we read team health instead" is a strong answer, the kind that closes the candidates you most want.

One honest note, because tooling can't carry this alone: no dashboard fixes burnout. Workloads and staffing do, plus the willingness to actually change something. The most a good signal can do is point that effort at the right team, early enough to matter, without asking anyone to trade privacy for help.

Try the flow on the live demo, or start free and run your first check this week.

Earlier in this series: your AI summary shouldn't require a copy of everything, the same keep-the-data-where-it-lives principle applied to security.