Menu
Search
8 June 2010

How to make a 2-week Agile sprint

One of the hardest parts of Agile development, but also one of the most powerful and rewarding, is figuring out how to make the process work for the team you’re part of.  Even though the guidelines are clear, there is simply no “one size fits all” approach when it comes to Agile.

We as Product Owners need to loosen up a little when it comes to pre-determined process, and work with our development teams to design an Agile process that works for everyone involved.  With this in mind, developer @darb and I evolved the following sprint guidelines for one of the applications we are responsible for here at Yola.

The ingredients of a good sprint

Every two weeks we sit down and discuss the next sprint, and we make sure the following ingredients are included:

  • 1 x silly name. Darb is partial to automatic Linux-generated names like Thundering Tuatara.  I am partial to non-silly names like “Sprint 3”.  But he won this argument — it’s all a game of give-and-take.
  • 1 x revenue-related. Every sprint has to include at least one bug fix, enhancement or feature that will improve our top-line revenue.  We are a business, after all, and this part is too often neglected by Product Owners.
  • 1 x front-end maintenance. With every sprint we have to make an improvement that customers will notice.  This can be small visual tweaks to increase usability, or bigger overhauls of the interface based on customer feedback and analytics.
  • 1 x back-end maintenance. Performance improvements and general back-end maintenance get put on the back burner way too easily – and as Product Owners we are often very guilty of deprioritizing maintenance because you can’t immediately see the effects of working on this.  But balance is extremely important, and paying down technical debt needs to be part of every sprint.
  • 1 x error messaging/UX improvement. This is related to front-end maintenance but different in that these tickets aim to address user experience and usability improvements mainly through changes in the interaction, not necessarily the visual design.

It’s important to note that these buckets aren’t mutually exclusive, so some tickets will address more than one issue — and that’s even better.  This is also the minimum requirement, usually we try to get a good balance of all kinds of tickets into a sprint.

The recipe for a good sprint

In addition to these minimum requirements of what should go into each sprint, we put together the following guidelines:

  • Leave time for ad hoc planning and work.  Other stuff will always come up.  Don’t let it make you miss your deadlines — leave room for it so that you have “planned outages” in your sprint work.
  • Make sure there is a good mix of big and small size/complexity tickets.  Momentum is so important.  Seth Godin talks often about the importance of shipping often, and to do that, you have to bite off manageable chunks of work.  Make that small text edit that’s been on the backlog for months because it’s just too small to schedule.  And couple that with bigger tasks that get you closer to your overall product vision.
  • Make effective use of priority. We use Jira to track our projects, and we use Priority to schedule the order of work.  We have to get P2 and P3 projects done in the sprint.  P4 and P5 projects are nice to have, and we do them if there is time.  For P1 work, see the first bullet…
  • Stir Rian into vitriol. I’m not sure why it is important for Darb to find new ways to freak me out every week, but he seems to thrive on it, so this remains in the recipe for now.
  • Simmer in QA by mid 2nd week.  It’s such a simple thing, but so many organizations forget that for something to release on Thursday, it doesn’t mean that first round development is completed by Thursday morning.  It means that at least some first round development needs to be completed by Tuesday morning so that QA can start.
  • Rinse and repeat.  And then we do it all over again.

The “ingredients” and recipe are very specific to the application that Darb is the lead developer on, which is our payment system — with all its many back-end and front-end complexities.  Obviously different applications need different ingredients.  On our website, for example, we place a lot more focus on front-end than back-end.

And I think that’s the point I’m trying to make — that the best product development process is the one that Product and Engineering agrees on, and it never looks exactly the same across all applications.