Why leave estimation at the story level if it sums up with sub tasks?

Pietros

A few days ago, I received the following answer:

 If you are creating a story on the spot, put in a rough estimate as a placeholder and then when you break down into sub-tasks you can remove the estimate from the story.  That being said, I recommend leaving the "original estimate" as the sum of the tasks and the remaining estimate (still at the story level) as 0 so you can get valid metrics and forecasting.

I do not think it is correct - when I leave an estimate at the story level and estimate sub-tasks, it will sum-up. In my opinion, there is nothing wrong with leaving an estimate to 0 on the story level and estimate sub-tasks. Or am I missing something?

Nic Brough -Adaptavist-

I'm not sure this is "correct" either.  I think JIRA is built to allow all the users to work in different ways.  Some of us need to break up stories into subtasks and estimate just them.   Some of us need to estimate subtasks and reduce estimates on their parent story as we go.  Some of us need to estimate a story and then use subtasks to add new things to the job at hand.  Some of us need to estimate a story and subtasks as individual pieces of work.

All of these approaches are supported in JIRA and it makes no effort to force you to do it in any one way.

What I suggest in all the questions about how estimate work is "talk to your users, work out what works best for you and go that way".  You are absolutely spot-on with "there is nothing wrong with leaving no estimate on a story and estimate the sub tasks" - it's a perfectly valid approach as long as your users know that is what you expect of them.  (It's also my preferred way to work - a story is a rough "something someone wants" and the subtasks are identifiable, discrete and estimate-able chunks of work that need doing to make it happen.  But that's just the way I do it when given free choice.  Other approaches work just as well, depending on the context)