Early in my career running people operations for a fully distributed company, I watched a 40-person startup install activity-tracking software after one slow quarter. Screenshots, keystroke counts, app usage scores — the works. Within six months, three of their most senior engineers had resigned, and every exit interview carried some version of the same sentence: they felt managed like children, so they started acting like contractors. Output did not recover until a year after the software was removed.
As an organizational psychologist who has designed remote onboarding for more than 500 new hires, I understand why leaders reach for surveillance. Distance feels like darkness, and when you cannot see people, monitoring software looks like light. But both the research and my own field work point the same direction: watching people measures anxiety, not contribution. The discomfort you feel as a manager is real, and the instinct to instrument something is reasonable — the mistake is aiming the instruments at people instead of the work.
This article maps the two operating models side by side, walks through the legal lines you should not cross, and gives you a transition plan you can defend to both your board and your team.
What the Data Says About Watching People
Start with where work actually happens. Gallup’s 2026 workplace research puts roughly 52 percent of remote-capable employees in hybrid arrangements, which means most teams now manage a permanent mix of in-room and remote people. Against that backdrop, 2026 workplace surveys report that 64 percent of employees say a strict return-to-office mandate or heavy monitoring would make them consider quitting. The people you most want to keep have the most options — and the strongest reaction to being watched.
Surveillance also fails on its own terms. It optimizes for visible activity, so rational employees produce visible activity: mouse movement, quick replies, a green status dot. The real work — thinking, designing, debugging, writing — becomes invisible by definition. You end up with a dashboard that proves compliance while actual output quietly drops, and the data you paid for tells you less than a weekly one-to-one would have.
Trust-Based vs Surveillance-Based Management
The two models are not variations of the same thing. They measure different objects, assign managers different jobs, and fail in different ways. The comparison below comes from patterns I have seen across dozens of distributed teams, not from theory.
| Dimension | Surveillance-based | Trust-based |
|---|---|---|
| What is measured | Hours online, keystrokes, app usage | Outcomes, delivered work, quality signals |
| Manager’s core job | Watch activity | Remove blockers and set direction |
| Employee response | Performative work | Ownership and early candor |
| How problems surface | Red flags on a dashboard | Missed commitments, raised early by the team |
| Talent effect | Pushes out senior people | Attracts and keeps senior people |
The last row is the expensive one. Monitoring is a tax you place on every honest employee in order to catch the few who might abuse freedom — and senior people, who always have options, pay that tax first by leaving. Trust-based management is not the absence of accountability; it is accountability attached to the work instead of the worker.
The Legal Lines You Should Not Cross
Even if you remain convinced that some monitoring is necessary, the law constrains how it is done. In the European Union, GDPR requires that monitoring be necessary and proportionate, that workers be informed, and in countries such as Germany that works councils be consulted before any monitoring tool is introduced. In the United States, states including New York, Connecticut, Delaware and California require employers to give written notice before electronically monitoring employees. Covert monitoring that feels routine to a manager can be unlawful on the day it ships.
Legal compliance is the floor, not the strategy. A policy buried in a contract does not build trust even when it is lawful. The teams I advise that use monitoring narrowly — for security posture rather than productivity scoring — write the policy in plain language, publish it to everyone, and name exactly what is never tracked. Transparency is what separates security operations from bossware, even when the underlying tooling looks similar.
Measure Systems, Not People
The alternative to watching people is instrumenting the system. Healthy remote teams track a small set of output signals that the team itself can see:
- Delivery signals: cycle time, sprint commitments met, change failure rate.
- Quality signals: production incidents, customer-reported issues, review turnaround time.
- Flow signals: pull requests merged per week, decision logs updated, handoff notes completed.
Notice what these share: they describe the work, not the worker. They are visible to everyone doing the work, so they help people find their own blockers instead of only reporting on them upward. Harvard Business Review research found remote workers are about 13 percent more productive than their office counterparts — but that advantage only survives in environments where energy goes into the work rather than into looking busy.
A useful test before adopting any metric: could a team member improve the number by doing something good, rather than by performing something fake? Keystroke counts fail this test in both directions. Cycle time mostly passes it. No metric is perfect, so run a small portfolio and watch for gaming behavior as a signal of a badly chosen metric, not a badly chosen team.
A Transition Plan You Can Defend
If you inherit a surveillance-heavy culture, do not switch everything off on a Friday afternoon and call it freedom. Sequence the change so people can see the replacement is real. Start with an honest inventory, then publish the new policy in words everyone can read:
What we track today: [list current monitoring honestly, including screenshots and idle time]. What we will stop tracking: [activity-level data, screenshots, idle minutes]. What we measure instead: [the delivery, quality, and flow signals above]. What we will never track: [keystrokes, private messages, personal devices].
Then run a 30-day pilot with one or two teams. Publish the output signals weekly, hold a retrospective at the end, and compare missed commitments — not hours — against the surveillance period. Leaders I have walked through this expect slackers to emerge. What usually emerges is the opposite: blockers that were invisible under surveillance become visible, because people finally believe that surfacing a problem will not be read as underperformance.
What Changes After You Stop Watching
The startup from my opening eventually made the switch. It took a quarter of visible consistency — leaders publishing their own metrics, managers asking about blockers before deadlines — before engineers stopped performing status and started reporting risk early. Delivery predictability improved. So did retention. Nothing about the tooling changed; the measurement target did.
Trust-based management is not soft. It is demanding in a different direction: clear outcomes, visible signals, and leaders who accept being measured by the same standards they apply. If your remote policy can name what it measures, publish it openly, and survive the question of whether you would accept it applied to yourself, it is probably built on trust. If it cannot, you are not running a productivity system — you are running an anxiety generator with a dashboard.