You are at a beautiful snowclad hill that has a drop of approximately 10 meters. The hill has an incline of approximately 35%. How long do you estimate it take for you to ride your sled to the base of the hill?
Most of us will probably give an answer of something in the rage of a few second. And that would probably be right, if you were at the top of the hill. But what if you started out at the bottom of the hill?
Ok, this may feel like a trick question. But these kind of assumptions is one of the reasons why estimation of features and user stories so often are wrong. We tend to only focus on the value adding part of what we are supposed to do. But just like in the sled ride, we often need to do a lot of non-value adding work to be able to do the value adding work.
If we want to know how long it will take to complete a feature or a user story it is much more accurate to use historical data. One big reason for this is that historical data will not only take the value adding time needed but it will include the non-value adding parts as well.
If you have a reasonable stable system of work, using historical lead-time data and probabilistic forecasting will be much more accurate and much cheaper to calculate.
This is my Lean/Agile Advent Calendar. I will publish a short post on a Lean/Agile topic every day up until Christmas. I will based each days topic on what is behind the door in the LEGO® City Advent Calendar. So be sure to check back every day!
DISCLAIMER: LEGO® is a trademark of the LEGO Group, which does not sponsor, authorize or endorse this post and blog in any way.