I Tracked Every Hour I Spent Typing for 30 Days. Here's What I Found.

It started as an argument with myself. I had a hunch that I was spending more time typing than I realized — that the keyboard was eating my day in five-minute increments I never noticed. So I did what every overcaffeinated knowledge worker does when they want to feel scientific: I built a tracker.

For 30 days, every keystroke I made was logged with a timestamp and the active application. I wrote a small launchd job on my Mac that captured key events at the OS level (no actual content, just counts and timestamps), bucketed them per app, and exported the totals nightly. The results were not what I expected. They were worse.

That number — 6.2 hours per workday — is what shocked me. I work roughly 9 hours a day. Two-thirds of my workday, I am hitting keys. The remaining third is reading, listening, thinking, or in meetings.

I am not a writer. I am not a transcriptionist. I work in product, which is supposed to be 70% talking and 30% doing. And yet, by the cold metric of "fingers moving across keys," I am a typist.

The Setup

The tracker logged three things every minute it detected activity:

1. Active app. Which application had focus when keys were pressed.
2. Key event count. How many keystrokes happened in that minute.
3. Active duration. How long I was actually typing versus pausing between bursts.

I defined "actively typing" as a 60-second window with at least 30 keystrokes. That filters out windows where I was just clicking around or scrolling. The thresholds are arbitrary, but applied consistently across the 30 days they give a fair relative picture.

I tracked 22 workdays (Monday to Friday, with two days off for travel) and 8 weekend days. Weekend numbers are lower but not by as much as I expected.

Where the Hours Went

Here is the breakdown, averaged across the 22 workdays:

The first thing that jumped out: three apps account for two-thirds of my typing time, and none of them are the apps I would have called "my main work." I would have told you my job was building product. By keystroke count, my job was Slack.

The Composition: What Was I Actually Typing?

I wanted to know whether the typing was thoughtful or repetitive. So I sampled five random one-hour windows from each week and manually classified what I had typed during them. (I had access to my own messages and docs, so this was straightforward.) Categories:

• Original composition — writing something new that required thinking
• Restating / paraphrasing — rewording something I had already said or read
• Acknowledgment — "got it", "thanks", "sounds good", "+1"
• Status / update — telling someone the state of something they asked about
• Code or technical input — actual code, queries, commands
• Filling forms — fields, selectors, names, dates

The breakdown across 20 sampled hours:

About 43% of my typing was either acknowledgments, status updates, or paraphrasing. None of that is hard cognitive work. It is the equivalent of saying "yeah, I'll handle it" 200 times a day, by typing it instead of saying it. If I had a way to convert speech to text reliably, in any app, that 43% would be the first thing to go.

The Things I Expected to See, That I Didn't

I expected to find that I was typing more in long, deep blocks — composing documents, writing PRDs, drafting emails. The data does not back this up.

The median typing burst was 54 seconds long. Half my typing happens in chunks shorter than a minute. Slack messages, email replies, comments on a Linear ticket, a quick Notion edit. The deep, hour-long writing sessions I imagined I was having? Those exist, but they are rare. Maybe two or three per workweek.

This matters because of what I will call the input mode tax. Every time you start typing, your brain has to switch from whatever it was doing — reading, listening, thinking — into "compose mode." Each switch has a cost. When you are typing in 54-second bursts, you are paying that cost dozens of times an hour.

The Math, Done Honestly

The "30,000 hours of your life spent typing" stat that gets thrown around in productivity blog posts is bait. But let me do the version of that math that survives scrutiny.

I type 6.2 hours per workday. That is 1,550 hours per year (250 workdays). Across a 40-year career, that is 62,000 hours typing. To put that in perspective: that is 7 full years, awake or asleep, of just typing. It is more time than most people spend on any single activity in their lives, including sleeping in any given decade.

If 43% of those hours go to acknowledgments, status updates, and paraphrasing — the low-cognitive stuff — that is 27,000 hours of low-value typing across a career. Three full years of life, dispatched in 54-second bursts, mostly saying "got it" with extra characters.

What Changed When I Started Dictating

For the second half of the experiment, I started dictating instead of typing whenever the task was a short reply, a status update, or a Slack message. I kept typing for code, deep composition, and editing.

I used Voice Keyboard Pro — disclosure: I work on it, that is partly why I ran this experiment. Hold a hotkey, speak, release, text appears at the cursor in any app. The thing that surprised me was not the speed. It was the cognitive shift.

When I type a Slack message, I compose it twice — once in my head, once with my fingers. The second pass is where I edit, hedge, and pad. Half my Slack messages are 30% longer than they need to be because typing gives my brain time to add filler. When I dictate, the message comes out closer to how I would say it out loud — shorter, more direct, often clearer.

Half my Slack volume disappeared. Not because I sent fewer messages, but because each message was shorter and more decisive. The recipients noticed.

The Numbers, After

For the second 15 days of the experiment:

• Active typing time dropped from 6.2 hours to 4.4 hours per workday
• Estimated text output stayed roughly the same (measured by character count of sent messages and committed work)
• The savings concentrated in Slack (-44%) and email (-38%)
• Code typing was unchanged. Document drafting was unchanged. The displaced typing was almost entirely in the "communication" category

1.8 hours per day. About 450 hours per year, or 18,000 hours over a career. Two full years of life given back, roughly.

Caveats

I want to be honest about what this experiment is and is not.

Sample size of one. This is my data, not a study. Your typing patterns may be very different. A developer who codes all day will see far less benefit than a manager who lives in Slack.

Self-selection bias. I work on a dictation tool, so I was probably more motivated than the average person to make it work.

The first week was rough. Dictation feels weird until it doesn't. Day one, it felt like talking to myself in public. By day five, I was forgetting I was using it. Most people quit before day five — that is the real reason dictation has not taken over.

The accuracy threshold matters. If transcription accuracy drops below about 95%, you spend more time correcting than you save. I would not have run this experiment with the dictation tools that were available in 2018.

What You Should Take Away

Even if you do not care about dictation, two things from this data are worth knowing:

You probably type more than you think. Most knowledge workers I have shown this data to underestimate their typing time by half. They picture the long deep-work sessions and forget the 200 small bursts. If you have been thinking your computer setup, ergonomics, or repetitive strain matters less than it does — it almost certainly matters more.

Most of your typing is not the part that matters. The original composition — the part where keyboard mastery actually pays off — is a third of your time at most. The other two-thirds is acknowledgments, status, and paraphrasing. That is the part where speed and accuracy matter least, and where any reasonable alternative would do.

I tracked 30 days of typing and found out I am a typist. Then I tracked 15 more days of dictating and found out I do not have to be.

Want to Run Your Own Experiment?

You do not need 30 days of data to learn something useful. Try this instead:

1. For one workday, every time you finish a Slack message or email, glance at the clock. Was it more than 60 seconds? More than 2 minutes? Note the rough duration and length.
2. At the end of the day, add it up. Most people are at 90 minutes to 2 hours just on Slack and email composition.
3. Try our calculators to see what that time would be at speaking speed instead: Time-Saved Calculator, Keyboard vs Voice Race.

You do not need to switch to dictation. You do need to know what you are choosing.