If you lead any kind of development then I’m talking to you, and I am telling you to just stop wasting time on guesswork estimations.
Humans are terrible at predicting things, especially about the future. But those who depend on us still want to know when we will deliver. So we guess, because not guessing is bad for business.
It is well known that we are better at comparing than predicting. With two tasks next to each other we can guess which one is the harder one, and by about how much. Also, the simpler the task, the more accurate our estimations.
We combine the two by choosing a small standard size for tasks. We whittle down each task into smaller parts until they share that same small size in the sweet spot of estimation. For us, that optimal size takes about half a day to complete.
We also allow simple to-do items that take no significant time, and time-box messy tasks of research or infrastructure to 2 days.
A few years ago, I reviewed all projects I had managed from the last six months. To my surprise, most projects were delayed by roughly the same amount, 25%. Why the consistency?
And then it hit me. There is no estimate. I have not done a single estimate in over 10 years. I no longer design projects that will take X days to complete. I design projects worth X days.
What I realized was this: We had always handled the important and high-risk aspects up-front. By the time we had run over by 25% and management was anxious to start the next thing, whatever remained was typically less valuable. So we shipped and moved on.
Stop pretending you can predict the future.
Divide your problems into small low-risk tasks.
Do the important parts first.
Ship when you are out of time.
Less theatre. More progress