Monday, April 11, 2011

How we do Sprint Zero

Today, a reader of my newsletter asked me,
What is the purpose of Sprint Zero?

A colleague of mine argued that during Sprint Zero, the team implements the most important features from the product backlog with the object of determining the team's velocity. I disagreed, because we have the velocity from previous sprints/projects. So could you help me to understand clearly what is Sprint 0 about and when and why do we need it? I really appreciate your helps and articles. Thank you!
Sprint Zero is often used to mean 'getting ready for the first sprint.' Some Scrum coaches don't like the term because every sprint should produce an increment of value for the customer or user. A sprint 0 which creates say, artifacts like a design document or development environment, violates this rule, because such artifacts have no value for the customer or user.

Still there are some things you need to have ready before you can do your first sprint planning:
  • You must have the roles identified (Product Owner, ScrumMaster, Team).
  • You must have an initial product backlog (and you probably should have a product vision). 
  • Enough of the product backlog must be estimated and prioritized that the team can pick a reasonable amount of high value work from the backlog in the first Sprint Planning.
  • You might need to train your Product Owner and Team on Scrum.
By the end of Sprint Planning 1 (first half of the sprint planning meeting), you will need to have defined:
  • the sprint length
  • the initial Definition of Done.
All of these are legitimate things to address during "Sprint Zero" or the time leading up to the first sprint planning.

What does not come into 'sprint zero'? Creating the development environment, test environment, high level design or other artifacts, and any development or testing. Actual work related to the project (as opposed to getting ready to do Scrum) gets prioritized in the Product Backlog, then is agreed upon during the Sprint Planning 1, and must produce some increment of customer value.

I wrote previous about my experience with a Sprint Zero. I hope the above and the blog entry are helpful!


Tricia Rosetty said...

Where I'm really struggling though is with the idea of a potentially shippable product as a result of sprint 1 as you describe it (creating the dev. environment, design groundwork). These aren't shippable products (i.e., initial development work without a client-friendly "thing to look at"), and waiting until they are would call for an extended sprint period (which isn't consistent with the idea of setting a regular period of time for sprints). How do you reconcile this when you're building something brand new?

Peter said...

Hi Tricia,

Great question, and I will pick up on that in blog entry, because the scope is too big for proper answer answer. But here's a hint anyway...

I would suggest you start with the most important functionality of the project, and shrink the acceptance criteria so small, that you can do it in one sprint.

For example, let's say it were 19-ninety something and you were starting to build "As a customer, I want to buy a book." How do we shrink that down so you can implement it in one sprint.

So the PO agrees with team: Show a web page with a book title (For Instance, Agile Project Management with Scrum, by Ken Schwaber). Show a 'buy' button. Click the button. See message "Congratulations you have bought.... It will be shipped to.... You will be invoiced at..." Everything, even your name and shipping address is predefined.

So from Sprint 1 on, "you" can buy a book. Then you start to enrich. Display more than one book. Search for books, set the shipping address, create a shopping cart.. etc.

I call this the "Chilean Rescue Model" of story design after the approach used to save the trapped miners.

Now, in your first sprint, you wont get much implemented, because setting up that dev environment takes some time. Hint: it is easier if you keep your teams together and can reuse your environment from the previous project.