Introduction
Every Sunday night at 9pm, I used to do something I called "time archaeology."
I'd open my laptop, stare at a blank spreadsheet, and try to reconstruct what I had done across four clients over the previous five days. Calendar. Slack history. Git commits. Browser history. Sometimes a sketchbook on my desk if I'd been wireframing.
It took 60-90 minutes. Every week. And the time I logged at the end of it was — at best — a guess.
This post is about what that guessing actually costs, and how I stopped doing it.
The Pain Point Most Freelancers Don't Name
If you log time at the end of the day or end of the week, you are not tracking time. You are reconstructing it. And reconstruction has three problems:
It's lossy. You forget the 15-minute call on Tuesday where you walked the client through a bug. You forget the 40 minutes of "just going to read these requirements before I start." Studies of self-reported time consistently find people under-report by 15-25% when reconstructing later.
It's biased toward big visible things. You remember the 4-hour deep-work block. You forget the seven 8-minute Slack interruptions that broke up your morning. The big block goes on the timesheet. The interruptions don't. Guess which one your client is paying for.
It takes time itself. That Sunday-night 90-minute ritual? That's 90 minutes you cannot bill anyone for, every single week. Over a year that's about 78 hours — two full work-weeks of unbillable archaeology.
Why I Couldn't Just "Try Harder to Remember"
For a long time I assumed this was a discipline problem. I'd promise myself I'd log as I went. By Wednesday I'd be behind. By Friday I'd given up and was back to Sunday-night archaeology.
The reason isn't laziness. It's friction. If logging a task takes 30 seconds of context-switch — open spreadsheet, find the right row, type the right description, mentally calculate the duration — you will not do it 40 times a day. Nobody will. Your brain protects itself.
The fix isn't more willpower. The fix is making the friction so low that not logging is harder than logging.
What Actually Worked
Three changes, in order of impact:
1. A timer that's always one click away
I switched to FillTheTimesheet, which keeps a timer button in the browser. One click starts a timer for the current task. One click stops it. The friction is genuinely below my mental friction floor — it's faster to start the timer than it is to not start it and feel guilty about it.
2. Pre-built project/task categories
Instead of typing "frontend work for Client X" every time, I built out a category tree once: Client X / Sprint 4 / Auth module, Client X / Code Review, Client X / Meetings. Now logging is a click into the right bucket — no typing.
This sounds trivial. It isn't. The category tree is the difference between accurate logs and "Client X — 4 hours."
3. Killing the Sunday ritual
This is the magic. Once real-time tracking became the default, Sunday night went from 90 minutes of archaeology to 5 minutes of review. I get back roughly 78 hours a year that used to evaporate into reconstruction.
How FillTheTimesheet Fits In
I'm biased — I built it partly because of this exact problem. But the principle is what matters: any tool where logging takes more than ~5 seconds will lose to procrastination. Pick the lowest-friction tool you can, and pre-build your project structure before you need it.
Key Takeaways
- "Logging time at the end of the week" is reconstruction, not tracking — and it leaks 15-25% of your hours
- The real fix is friction, not discipline
- Pre-build project/task categories so logging is one click, not typing
- Kill the Sunday-night ritual; that's two work-weeks a year you can't bill for
- Track in real time and your records become defensible documentation, not best-effort guesses
If this resonates, I write more freelancing and engineering essays as The Speed Engineer on Medium.