Why your estimates always come up short
Psychologists call it the planning fallacy: the tendency to underestimate how long your own tasks will take. Daniel Kahneman and Amos Tversky named it in the late 1970s, and it has turned up reliably in research ever since.
The study everyone cites asked Canadian undergraduates in the 1990s how long their honours thesis would take. The average guess was about 34 days. The average reality was around 55, and well under half finished by their own predicted date. The striking detail is that even when students were asked for a worst-case estimate, a good number of them still missed it.
The usual explanation is that when we estimate, we replay a mental movie of the work going smoothly. The half hour lost hunting for a file, the bug nobody expected, the colleague who needs something now — none of it is in the movie. And when we remember that a similar job overran last time, we explain it away as a special case and leave it out of this estimate.
Estimate in sessions, not hours
Francesco Cirillo built estimation into the Pomodoro Technique from the start. At the beginning of the day you write next to each task how many Pomodoros you expect it to take; at the end of the day you write down how many it actually took.
Counting sessions instead of hours is not just a quirk. An estimate of "three hours" quietly absorbs breaks, replies and staring out of the window. A session count measures only focused blocks. That puts estimate and outcome in the same unit, so you can compare them honestly and see exactly where the gap opened up.
How small to cut
Cirillo's rule of thumb is simple. Anything over five to seven Pomodoros is too big to estimate well, so split it. Anything smaller than one Pomodoro gets batched with other small tasks into a single session.
Each piece should be something you can clearly call done or not done. "Work on the report" has no end. "Fill in the three tables in section two" does. The sharper the finish line, the better the estimate.
Comparing estimates with reality
Estimating on its own does not cure the planning fallacy. Comparison does. For one week, write the estimate and the actual count side by side for every task. For most people the actuals come out larger by a fairly consistent ratio.
- Next to each actual count, note the reason for any big miss in a word or two: "research", "bug", "interrupted".
- After a week, divide total actuals by total estimates. If the answer is 1.5, multiplying future estimates by 1.5 is the simplest correction you can make.
- If the same reason keeps appearing, give it its own line in the estimate. If research always costs an extra session, write "research: 1" up front.
- Track different kinds of work separately. Writing and problem sets rarely miss by the same ratio.
Think about similar jobs, not this job
One of the more robust remedies in planning research is to stop starting from the details of the task in front of you and start from how long similar tasks actually took in the past. It is called reference class forecasting, and it is used for everything from household chores to major infrastructure projects.
Once you have a few weeks of session records, you can do this for free. Look up how many sessions the last term paper or the last feature took, then decide only whether this one is bigger or smaller. Your log is a far more honest witness than your imagination.
Putting it into a daily plan
The most common mistake when turning estimates into a day is filling every available session. If you can realistically run ten 25-minute sessions, a ten-session plan is almost guaranteed to slip. Plan 70 to 80 percent of your capacity and leave the rest open. The empty slots are where overruns land.
At the end of the day, carry unfinished tasks forward and re-estimate them. Knowing how many sessions have already gone in makes the remaining estimate far more realistic. This timer shows your completed session count on screen, so the whole record is just copying that number next to your estimate before you close the tab.
Expect your estimates to be wrong for the first few weeks. The goal is not to hit them on the first try, but to learn how far off you usually are and in which direction. Knowing that alone cuts down the number of things you were sure would be done by tonight.