A colleague of mine asked me "how do you create new story point estimates out of old estimates if the old estimates were wrong?" This is a good question for Scrum teams to consider as part of continuous improvement, how do we ensure we keep getting better at Story Point estimating? Here is one approach.
When you start your sprint your product backlog items (PBIs) have been story pointed, and sometimes during the sprint you realise that a PBI has been incorrectly story pointed. Don't be tempted to change the story points on a PBI during the sprint as that will mess up your tracking and burndown charts. If you have a PBI that is "ballooning" put some kind of visual reminder on the PBI in the sprint backlog so you can identify it later. Like a sticker of a balloon for example.
Here is the important bit.
When your sprint is over and hopefully that ballooned PBI is sitting in the Done column, its time to clear down your Sprint Backlog (SB). Just before you clear down the look at your Done PBIs. For the ones that ballooned, re-story point them to what they actually should of been so that you can learn from it. Do the same for any PBI that is over or under what it was originally story pointed at.
I usually then go a step further and suggest sticking good examples of a 3 story point PBI and a 5 story point PBI up on the wall near the Sprint Backlog to form a bit of a story point estimate library to help when using triangulation. The Product Owner & Team members can point to it with new PBIs in hand and ask"is it bigger or smaller than that one?"
This story point estimate library allows us to learn from our bad estimates and ensure that we continue to use triangulation estimating based on good foundations.