Remote Employee Productivity: What to Measure and What to Ignore
Activity percentage, hours logged, and messages sent are the three worst productivity metrics in common use. What to track instead when you cannot see people work.
When a manager loses line of sight, the instinct is to find a number that replaces it. The numbers that are easiest to collect are almost exactly the numbers that mislead you most.
The three worst metrics, and why they fail
Activity percentage
Derived from keyboard and mouse input. It answers one question well — was this machine clocked in and untouched for three hours? — and nothing else.
It cannot distinguish a developer reading an architecture doc from a developer reading the news. It rates data entry above analysis. Rank your team on it and you systematically punish the people doing the hardest thinking, while anyone who wants to game it can do so with a $12 mouse jiggler.
Use it as an anomaly detector. Never as a scorecard.
Hours logged
Hours are a payroll input. Treating them as a productivity measure creates a direct incentive to work slowly, and it punishes exactly the person you want more of — the one who finishes in six hours what takes others nine.
You still need accurate hours. You need them for payroll, overtime compliance, and job costing. Just do not confuse the input with the outcome.
Communication volume
Messages sent, meetings attended, response time. This measures visibility, not contribution. Reward it and you get a team that is extremely responsive and produces very little, because every deep work block is interrupted by the performance of availability.
What actually predicts performance
Throughput on a stable unit. Tickets closed, jobs completed, invoices processed, estimates delivered. Not perfect, but tied to reality, and the distortions are visible — if someone games it by cherry-picking easy tickets, that shows up quickly.
Cycle time. How long from start to done. Rising cycle time is the earliest reliable signal that someone is stuck, overloaded, or blocked — usually weeks before it shows up in output volume.
Rework rate. Work that comes back: corrections, callbacks, revisions. Fast output with high rework is not productivity.
Commitment reliability. Did the thing they said would be done Thursday get done Thursday? This is the most useful single signal in remote work, and it requires no software at all.
Their own read. Ask in the one-on-one: what slowed you down this week? People know. The information is free and nobody collects it.
Where hours data does belong
None of the above means hours are useless — they answer different questions:
- Payroll and overtime. You are legally required to get this right, across all hours worked in the workweek regardless of where they happened.
- Job costing. What did this project actually cost in labor? You cannot price the next one without it.
- Capacity. Is this person at 30 hours or 55? Both are problems, and you cannot see either without a record.
- Client billing. If you bill hourly, this is not optional.
That last-but-one is the one managers most often miss. Remote overwork is invisible by default — nobody sees the person still at their desk at 9 PM. An accurate hours record is one of the few ways to catch it before it becomes a resignation.
A workable weekly rhythm
Weekly: look at throughput and cycle time by person. Look for changes, not absolute levels — people work at different speeds and that is fine.
Weekly one-on-one: ask what got in the way. Do not open with metrics; open with the question.
Monthly: look at hours in aggregate. Anyone consistently over 45 or under 30 gets a conversation, and in both directions it is a support conversation, not a discipline one.
Never: rank the team on activity percentage, or bring up a specific hour on a specific day unless there is a concrete payroll dispute.
What Klees Desk gives you, and what it deliberately does not
Klees Desk records hours accurately for staff on Mac and Windows: clock in and out from the menu bar, idle time surfaced as a question rather than a silent decision, and hours that flow into the same timesheets and payroll export as your field crews.
It reports an activity level, because it is genuinely useful for spotting a machine left clocked in. It does not compute a productivity score, and it does not rank employees against each other — because that number would be wrong and acting on it would cost you people.
It never logs keystrokes, records nothing while someone is clocked out or on break, and shows every employee the same view of their own data that their manager sees.
Related reading
Compliance and payroll lead at Klees. 15 years in construction payroll, prevailing wage, certified payroll, and OSHA reporting. CPP certified.
Related reads
How to Manage Remote Employees Without Micromanaging Them
The practical difference between oversight and micromanagement, and how to get visibility into remote work without destroying the trust that makes remote work function.
Is Employee Monitoring Legal? A US Employer's Guide
What US law actually requires before you monitor remote employees — notice rules, state-by-state differences, and the practices that create real legal exposure.
How to Write a Remote Work Monitoring Policy Your Team Accepts
A monitoring policy people actually read, with the six sections that matter and language you can adapt — plus the mistakes that turn a policy into a resignation letter.