Agile Principle #7: "Done" Means "DONE!"
by Kelly Waters
In agile development, "done" should really mean "DONE!".Features developed within an iteration (Sprint in Scrum), should be 100% complete by the end of the Sprint.
Too often in software development, "done" doesn't really mean "DONE!". It doesn't mean tested. It doesn't necessarily mean styled. And it certainly doesn't usually mean accepted by the product owner. It just means developed.
In an ideal situation, each iteration or Sprint should lead to a release of the product. Certainly that's the case on BAU (Business As Usual) changes to existing products. On projects it's not feasible to do a release after every Sprint, however completing each feature in turn enables a very precise view of progress and how far complete the overall project really is or isn't.
So, in agile development, make sure that each feature is fully developed, tested, styled, and accepted by the product owner before counting it as "DONE!". And if there's any doubt about what activities should or shouldn't be completed within the Sprint for each feature, "DONE!" should mean shippable.
The feature may rely on other features being completed before the product could really be shipped. But the feature on its own merit should be shippable. So if you're ever unsure if a feature is 'done enough', ask one simple question: "Is this feature ready to be shipped?".
It's also important to really complete each feature before moving on to the next...
Of course multiple features can be developed in parallel in a team situation. But within the work of each developer, do not move on to a new feature until the last one is shippable. This is important to ensure the overall product is in a shippable state at the end of the Sprint, not in a state where multiple features are 90% complete or untested, as is more usual in traditional development projects.
In agile development, "done" really should mean "DONE!".
See also:
10 Key Principles of Agile Software Development
Time waits for no man!
How d'you eat an elephant?
Fast but not so furious!
8 April 2007 21:24
I agree entirely. One way I have heard it compared is to say, there is "done" and then there is "done done done." The difference is that the former might be when you feel that you are finished working on it, but "done done done" means it has been reviewed (both design review and code review), it has been tested (by yourself and hopefully a QA team) and it is ready to release in every possible sense of the term.
8 April 2007 23:32
I certainly agree with your posting, but as I posted on here, I see a large problem getting some features to shippable in a short amount of time.
If you have 1 user story that takes the team 2 sprints to complete you surely can't be done at the end of 1 sprint. And if the feature isn't shippable till say 5 user stories are complete then you may be able to only get to 2 of them per sprint, and you wouldn't be shippable till sprint 3.
I'd be interested on your thoughts.
9 April 2007 17:04
First of all, you can cheat a bit :-)
In my mind, a user story equals a feature, so in the case where there are 5 user stories, treat as 5 features.
If the product isn't shippable until all 5 features are complete, because it doesn't really make sense piecemeal, that's fine.
It's the feature that has to be shippable quality; not necessarily that it would be appropriate to ship the product after every feature.
But, when you say a feature is complete, it must really be shippable once the dependent features are also complete.
With features broken down into their smallest possible pieces, if you still have 1 feature that spans 2 sprints, you should seriously consider whether your sprints are too short for your situation.
Your product, tools, team size, skills, features, etc, etc determine what is an optimum sprint length for *you*
Hope this helps...
Kelly.
9 April 2007 17:09
If you want to discuss this issue or others with your peers, why not give the forum a try...
http://www.groups.google.com/group/allaboutagile
10 April 2007 21:21
Fine for software development, but I think Google and 37 signals would differ on this, it needs to be still bug free and done, but not 100% done, if Google kept Gmail in secret until its out of "beta" it wouldn't be as much of a success.
14 April 2007 07:13
I agree. You don't have to develop all the intended features. You don't even have to make the features as functionally rich as you intend to. But you do need to make sure that the features developed in *each* Sprint/iteration are "DONE!", including testing, styling, user acceptance, etc.
28 October 2007 11:30
Our problem is the developers lacking experience "out in the field" doesn't really understand the effect of what they conceive as a minor problem or glitch. While other developers dive into details. tiny details.
8 January 2008 05:33
I agree that there is a need to make 'done' mean that "it" has completed its analyse to test cycle but I don't think it means the feature is done. I like the idea of scale (see Jeff Patton) where a feature can be split into separate deliverable independently useful parts. I.e. you can deliver the essentials of a feature in incr'n so that you could consider it usable. However, you still intend to improve it in incr'n+1. (I don't believe that these are typically new stories).
10 January 2008 17:43
What happens when you complete your stories near the end of the iteration and it will take QA several days to verify the story is complete and working? In the meantime you are planning for the next iteration and have no idea if you really are complete, until QA catch up all the time?
9 August 2008 05:48
I have exactly the same scenario in my team(As mentioned by anonymous reader above). Currently, we moved testing to next sprint. But - what if the testing team finds out a lots of bugs by the end of testing? Shud the bug fixing be moved to another sprint? Overall extending it to 3 sprints!!
29 August 2008 22:40
You shouldn't leave several days of testing until the end of the iteration. The testing for each task/deliverable within an iteration should be done as soon as possible (start testing on the first day if you can). As you approach the end of your iteration, you should already have everything tested and all bugs fixed.
21 October 2009 11:56
I'm very interested in your point "complete each feature before moving on to the next...".
We are building a shared agile development team, that is we have multiple projects with multiple project owners. I can imagine situation that the team completes an underestimated feature. This causes another feature to be shifted into the next iteration.
This is ok when this is in same project. But what to do when this affects another project in the iteration? That means finishing of a task in ProjectA becomes longer and there is no time for start the work on a task in ProjectB. How can I advocate this situation for ProjectB owner.
By the way I found only one article about How To Share An Agile Development Team. Do you know any other sources?