I've been building Timex for a while, which means I've been running it on my own workday longer than anyone. But there's a difference between shipping a time tracker and actually reading your own record, every evening, for a month. So I did that — 30 days of my own second-by-second data, the way a user would see it.

I expected the data to confirm what I already believed about my day. It didn't. It mostly corrected me. Five things, in the order they surprised me.

Short answer: a month of real time data surprises you in the gaps, not the apps: away-time turns out to be the story, "idle" hours are often media you meant to count, website rules change the past, and — the biggest one — you work more than your memory says you do. The record is kinder than your guilt.

1. The away gaps were the real story

I expected to learn about my apps. Instead the most interesting hours of my day were the ones with no app at all.

Timex stops counting after 2 minutes of no input — that's the idle gate, and it's what keeps the record honest. Then after 5 minutes away it offers the gap back: "Away 25 min · 10:05–10:30 → Add time…" Every evening I'd hit these prompts and think "oh right, the call with the client" or "the conversation in the hallway that turned into a design decision."

Here's what surprised me: my away-time was consistently longer than my memory of it. The "quick call" was never quick. The meetings I remembered as 15 minutes kept being 40. Not once in 30 days did I look at an away gap and think "that felt longer than it was." It was always the opposite direction. If you bill for your time, that gap between remembered and actual is money — and it only ever flows one way.

2. Media hours are honest hours

The 1.3.0 release added a rule I was skeptical about: audio playback or microphone use keeps tracking alive through the idle gate. Watch a long video, take a call with the mic on, and the time counts instead of evaporating into "idle."

Reading my own month, I get it now. There's a real category of work that looks like "doing nothing at the keyboard" — watching a conference talk with a notebook open, listening to documentation while sketching, a video call where you're mostly listening. The idle gate alone would erase all of it. The media exception keeps it, and my month was full of it. Without that rule I'd have lost a chunk of real working hours every single day.

The flip side exists too: sometimes "watching a video" is just watching a video. That's fine — the record shows the window title, and you decide what it meant. The data doesn't judge. It just refuses to pretend the hour didn't happen.

3. The browser was one big blur until I wrote rules

For the first days, my Chrome row was a single enormous number. "Chrome: 6 hours." Completely useless — those hours contained documentation, code review, client portals, and yes, some YouTube. The app can only record what it sees, and what it sees from a browser without rules is one gray box.

Website rules fixed it, and this is the part I'd underline for anyone starting: github.com → focus. youtube.com → its own category. The client's portal → its own category. A few minutes of rule-writing and the biggest app in my day split into real activities. The surprise wasn't that rules work — it was how much of my day the browser was hiding. Every tracker with per-site data says this. Living it is different: the blur didn't just hide distraction. It hid work.

And because recategorizing updates history, the rules I wrote in week two retroactively cleaned week one. The past became more accurate as I went. That's the correct behavior — your judgment about what counts shouldn't be frozen at install time — but living with it is odd at first, in a good way.

4. The idle gate earned my trust

I expected to fight the idle gate. "What if it stops counting while I'm thinking?" I've heard that objection a hundred times from users; I had it myself.

A month in: it never once bit me. Thinking at the keyboard involves small movements — typing notes, scrolling, clicking — and the gate only closes after 2 full minutes of nothing. What it actually caught was the truth: lunch at the desk with a video paused, walking away mid-paragraph, the 40-minute errand I'd forgotten. Hours that a naive tracker would have billed to whatever app was focused when I stood up.

That's the whole value of a record: it has to survive your skepticism. I stopped auditing the idle gaps by week two. The data had earned it.

5. My memory was unkind to me — the record wasn't

This is the one I want everyone who's ever felt bad about their week to hear.

Going in, I expected the data to confirm the guilt narrative: too much distraction, not enough deep work, the usual. Instead the record kept showing me working more than I remembered. The "lost" afternoon was in the data, focused, three hours deep. The "lazy" morning was a call, two away-gaps I'd logged, and an hour of documentation. The days that felt wasted mostly weren't — they were fragmented, which is a different problem with different fixes.

I'm not saying the data showed a saint. It showed real distraction too, honestly labeled, in the exact amounts. But my memory's version of my day was systematically crueler than the data's version. Thirty days of that correction changes how you talk to yourself about your work. I didn't expect a time tracker to be the kindest piece of software I own — but there it is.

How to do a time audit: what I'd tell someone starting their own 30 days

Don't buy anything first. Run the trial — Timex gives you 100 hours of real tracking, about two and a half work weeks — and do one thing: read the timeline every evening. Don't score yourself, don't optimize, don't write rules on day one. Just look. For a baseline to compare against, paint one day from memory in the free time audit worksheet before you start.

Somewhere around week two you'll hit the moment: you'll see a day you remember one way and the record shows another. That moment is the entire product. Everything else — the score, the rules, the exports — is downstream of deciding you'd rather know than guess.

Download Timex and start your own 30 days. Your record will disagree with your memory somewhere. I'd genuinely like to know where.

— DK