Tracking Time Against Measuring People

Time data can answer two very different questions. Where did the work go? And: is this person working hard enough?

The same records serve both. The purposes pull in opposite directions, and which one people believe is in play determines whether the data is worth having at all.

For a vendor-side perspective, Monitask also has a page on employee time clock software.

The two purposes

Understanding the work. How long things actually take, where effort concentrates, which projects consume more than they return, what to quote next time. The subject is the work.

Evaluating the person. Hours present, output per hour, comparison against colleagues. The subject is the individual.

The first produces useful data. The second produces defensive data. And the second contaminates the first, because people cannot record honestly for one purpose while being judged on the other.

Why measuring people breaks the numbers

Not a moral claim. A mechanical one.

People record what is safe rather than what happened. Gaps get filled, interruptions get absorbed into billable work, and the long unproductive afternoon becomes four neat hours.

Waiting disappears. Time spent blocked on someone else looks like idleness if idleness is penalised, so it stops being recorded — and waiting is the single most useful thing time data can reveal.

The difficult work looks worse. Anything with uncertainty takes longer, so the person doing the hard problem posts worse numbers than the person doing the routine one.

And once it starts, it does not stop. The records are technically complete and factually wrong, and no amount of tooling recovers them.

What honest time data is actually good for

Worth listing, because the case for tracking gets lost in the argument about surveillance.

Estimating. Your own historical figures are the only estimate source that reflects how you actually work. See estimating from your own data.

Pricing. Knowing what a piece of work costs to deliver is the basis of not losing money on it.

Finding where time goes that nobody intended. Meetings, context switching, rework, waiting on approvals. Usually surprising.

Deciding what to stop doing. A client or a project consuming three times what it returns is invisible without the data.

Billing, where work is billed by time.

And protecting people. Records that show consistent overwork are evidence, and they are evidence in the worker's favour more often than not.

The line

Simple to state and requiring discipline to hold.

Track what work was done and for how long. That is production data.

Do not track how the person behaved. Screenshots, keystroke counts, activity scores, application monitoring, idle detection. That is a different category with different consequences.

Do not compare individuals on time data unless the work is genuinely comparable, which it rarely is.

Do not use time records in a performance conversation if you told people they were for estimating.

That last one matters most. Using data for a purpose other than the one stated is the single act that converts honest records into defensive ones, permanently.

See what time data must never be used for.

The monitoring software question

Products exist that capture screenshots, log keystrokes, score activity and infer productivity. They are marketed as time tracking and are a different thing.

What they measure is presence and motion, not work. Someone thinking is idle by these measures; someone moving a mouse is productive.

They are legally constrained in many places, with notice, consent or consultation requirements that ordinary timekeeping does not carry. Rules differ substantially by country — check what applies where your people are.

They change the relationship. A team that is monitored behaves like a monitored team, and the behaviour includes producing the appearance the system rewards.

And they contaminate whatever else you were tracking. Once the timesheet is in the same system as the screenshots, the timesheet is a surveillance instrument.

See the line between timekeeping and surveillance.

How to tell which one you are doing

Four questions, honestly answered.

What will you do with a number you do not like? Ask why the work took that long, or ask the person to explain themselves? The answers describe different systems.

Would you show the team the whole dataset? If not, you are measuring people.

Is the reporting unit the work or the person? A report by project is one thing. A leaderboard is another. For broader independent background, see NIST Privacy Framework.

Did you tell them the purpose, and does the use match? This is the one that determines whether anyone records honestly.

Making it work

Say the purpose plainly, and then hold to it. Not once at rollout — whenever the data is used.

Report at the level of work, not the individual, wherever possible.

Make recording non-productive time safe. Waiting, blocked, rework, admin. If these cost the person something, they will not be recorded, and they are the most valuable thing in the dataset.

Show people their own data, and the reports made from it. Data that flows upward and never returns is data nobody has a reason to get right.

Act on what it shows. If waiting is recorded for six months and nothing changes, recording stops.

And accept lower precision. Honest approximate data is worth far more than precise fiction. See why time data is almost always wrong.

For people working alone

The distinction still applies, to yourself.

Tracking to learn how long things take is useful and it makes you better at quoting.

Tracking to judge whether you worked hard enough produces the same defensive behaviour, aimed inward, plus a fairly reliable route to feeling inadequate.

Track the work. Look at the patterns. Do not score yourself by the hour.

See time tracking for freelancers.

The short version

Two purposes: understanding the work, and evaluating the person. They pull in opposite directions.

Measuring people produces defensive data, and the contamination is permanent.

Waiting time is the most valuable thing in the dataset and the first thing to disappear when recording is risky.

Using the data for a purpose other than the one you stated is the single act that ruins it.

And honest approximate data beats precise fiction — accept the lower precision.