Why your blocks are too short, every single time

Not sometimes. Systematically, in one direction. And knowing about the bias does not fix it, which is exactly why the correction has to be mechanical.

The UnaTask team, Building UnaTask. Aug 21, 2026. 5 min read.

You blocked an hour. It took two. You are mildly annoyed at yourself, you assume today was unusual, and tomorrow you will block an hour for the next one.

It was not unusual, and there is a clean demonstration of why, from 1994.

Thirty seven students and their own thesis

Buehler, Griffin and Ross surveyed thirty seven psychology students about their honours thesis. Final semester, on a project they knew intimately and cared about a great deal. If anyone was going to estimate well, it was these people on this task.

They predicted an average of 33.9 days. The actual average was 55.5 days, about 64 percent longer.

The detail that makes it useful is what happened when the researchers pushed. Students were also asked for a best case, if everything went as well as it possibly could: 27.4 days. And a worst case, if everything went as poorly as it possibly could: 48.6 days. Reality came in at 55.5.

The average outcome was worse than their own worst case. Not the average case. The worst one they could imagine.

And when asked for a date they were 99 percent confident of hitting, only about 30 percent of them made it.

Three corrections that actually work

The reason this needs mechanical correction rather than effort is that the bias survives knowing about it. You have just read the study. You will still argue the next estimate down, because this particular task is different, and it always is.

Use the outside view, not a better inside one. Do not re examine the task and think harder about it. That is the inside view, and it is what produced the bad number. Ask instead: how long did the last three things like this actually take? The question is boring and far more accurate. It does require you to record actuals, which takes ten seconds at the end of a block.
Apply a fixed multiplier, and do not negotiate it. Start at 1.5 on your first instinct and adjust after a month of real data. Your gut says an hour, you block ninety minutes. Every time, with no case by case debate, because the debate is where the bias lives.
Split until a piece genuinely fits. If a task will not fit in a block, the problem is usually the task. "Write the report" is a project wearing a task's clothes. Cut it until you can picture the first sentence you would type. That is how you know it is small enough to start, which is a different and more useful test than whether it sounds small.

The part that compounds

Underestimation does not break one block. It pushes every block after it. A thirty percent error at nine in the morning becomes an entirely fictional afternoon, and by three you are not running a plan, you are running an apology.

That is the real argument for keeping your day only about sixty percent blocked. The unblocked forty is not spare capacity you failed to use. It is where the error gets absorbed instead of cascading through everything you scheduled after it.

This is one chapter of Blocks No. 01, a 26 page guide to time blocking. Free PDF, no email, and every number in it traced to the paper it came from. Get the full guide