Time tracking your team won't quietly sabotage

Nobody refuses to track time. They just do it on Friday at 4:50pm, from memory, in round numbers, against whatever project code they remember. The timesheet is complete, everyone's compliant, and the data is worthless.

That's the actual failure mode in most agencies — not rebellion, quiet approximation. And it's expensive twice over: you lose hours that were real and billable, and you lose the ability to know which projects make money.

What the approximation costs

Take a hypothetical five-person delivery team where each person, reconstructing the week on Friday, loses two hours they genuinely worked and could have billed. Small stuff: the 40-minute call, the revision on Tuesday, the hour spent unpicking a client's feedback.

5 people × 2 hours = 10 hours a week
10 × $150 = $1,500 a week
× 46 working weeks = $69,000 a year

That's an upper bound — it assumes every lost hour was billable and would have been accepted on an invoice. Halve it and it's still more than most agencies spend on software in a decade.

The second cost is worse and harder to see. If time entries are approximate, so is every project margin you calculate from them, which means your pricing is built on fiction.

Why people sabotage it, quietly

Not laziness. Four specific reasons, in rough order of importance:

They think it's surveillance. If time data has ever been used to question someone's effort, it's now a performance metric, and people manage metrics. Hours get logged to look right rather than to be right. This is the single most destructive thing an owner can do to their own costing data, and it usually happens by accident in one careless comment.

It's tedious out of proportion to the work. Fifteen fields, a project code they have to look up, a mandatory description, a dropdown of 40 tasks. If logging an hour takes ninety seconds, people batch it, and batching means guessing.

Nothing ever comes back. They fill in timesheets and never hear another word. No project ever changed because of the data. Why would they be careful?

The categories don't match reality. There's no code for "unpicking the client's feedback" or "waiting for assets", so it goes somewhere approximate — usually whatever's at the top of the list.

Design the system so accuracy is the easy path

Same day, always. Not real-time — same day. Real-time tracking demands someone start and stop a timer around every context switch, and agency work is nothing but context switches. A five-minute habit at the end of the day, while memory is intact, captures nearly everything.

Fewer than three fields. Project, hours, one line of description. That's it. Anything more granular is a tax you're paying in accuracy.

Fifteen-minute granularity, not six. Nobody's memory is accurate to six minutes, so six-minute increments produce precise-looking invented numbers. Fifteen is honest and fast.

Make non-billable a real, respectable option. Internal meetings, business development, training, admin, onboarding, dead time between projects. If there's no honest place to put those hours, they get absorbed into client codes and your project margins silently improve on paper. Non-billable time being visible is a feature — it's how you find out that a "small" retainer is consuming twelve hours a month of account management.

Pre-fill the obvious. Recurring meetings, standing retainer blocks, whatever happens every week. Reduce the thing to editing rather than composing.

Never make it retroactive-only. If someone can't fix Tuesday's entry on Thursday, they'll stop bothering.

Frame it as costing, not monitoring

Say this out loud, at a team meeting, in these terms:

This data is how we price work and decide what to say no to. It is not how I judge whether you're working hard. If your utilisation is low, that's my problem to fix, not yours to hide.

Then behave accordingly, because the first time you don't, the whole thing reverts. Concretely, that means:

• Don't use individual utilisation as a performance measure. Look at it by team and by project. If you need to discuss one person's hours, discuss the cause — assignment, scheduling, a stuck project.
• Don't react to a single high or low week. Weekly noise is enormous. Trailing four weeks or nothing.
• Never celebrate an over-utilised person publicly. You've just told everyone what gets rewarded, and you'll get it, at the price of a resignation next spring.

Close the loop, visibly

The strongest thing you can do to improve data quality is show people what it produced. Once a month, five minutes:

• "This project came in 30% over estimate. Here's where the time went, and here's what we're quoting next time."
• "Retainer X takes 14 hours a month and we priced for 9. We're renegotiating."
• "We turned down that RFP because we could see we didn't have the capacity."

Now the timesheet is an instrument the team can see working. That changes behaviour more reliably than any reminder or policy.

The corollary: if you're not going to look at the data, don't collect it. A tracking habit that feeds nothing is pure overhead and everyone can tell.

Handle the awkward cases explicitly

Fixed-price projects. People assume hours don't matter because the price is set. The opposite is true — fixed price is exactly where you can't see an overrun any other way. Say this every time.

Internal and pitch work. Track it. It's a real cost of winning business, and it's the reason your utilisation target isn't 90%.

Your own time. If the owner doesn't track, nobody believes it matters. This is the single most common reason time tracking fails at small agencies, and it's entirely within your control.

Overtime. If someone logs 55 hours, the correct response is to fix the staffing, not to question the entry. React badly once and they'll log 40 forever and you'll never see the problem again.

One change this week

Move the entry point in the day. Ask everyone to log time before they close the laptop, not on Friday. Say why, in one message: same-day entries are the difference between costing and guessing.

Then, three weeks later, take one project, pull the actual hours, and show the team what the margin turned out to be. That single act of showing people the output is worth more than any amount of chasing.