Tea-Driven Development » kanban

Fine-Slicing Beats Estimation for Predictability

Matt — Fri, 23 Sep 2011 09:42:46 +0000

As requested by JB in the comments to my previous post, here is some data about what happens when a team choose fine-slicing over estimation.

You’re about to see a CFD chart drawn by a team who used BDD to break down every requirement into scenarios before they started hacking on them. The items on the left aren’t actually scenarios in this case, they’re very small user stories which tended to be of a size of about 3-4 scenarios each. The point is, we broke everything down into the smallest pieces of behaviour we could, then re-assembled them into chunks that were meaningful enough to build together.

Rather than using story points to manage the variation in size of stories, we gave each story a value of one point, and used BDD analysis to try to ensure each story was a uniformly small size.

This data was collected over a period of about six months by a team of about eight developers. Their system (a high-volume website) was already live and they were adding features to it.

What strikes me the most is how straight the ‘done’ line is.

Slides from XP Day Talk

Matt — Sun, 14 Dec 2008 14:51:11 +0000

I’m just back from this year’s XP Day, London. Thanks to everyone who came and packed out the room to hear Rob and I talking about our experiences evolving our team from Scrum to Kanban. The slides are here.

There’s also a great transcript of the talk here on Tom Hume’s blog. Thanks Tom!

Kanban for Software Explained

Matt — Sun, 09 Dec 2007 22:55:31 +0000

Karl Scotland has posted a great description of how his team solved some issues they were having within their Scrum team by moving over to using a lean-thinking or Kanban system, based on a short buffer or Queue of Minimum Marketable Features (MMFs). It’s probably the clearest explanation I’ve seen yet of why and how to employ this emerging technique, and Karl certainly makes a compelling case for considering this as a progression for teams who are experienced with Scrum and need to be able to adapt rapidly during the development of a story or feature.

One of my key questions about Kanban is how it’s possible to predict long-term delivery dates for specific features, and although Karl goes some way to answering those questions, it looks as though you need, as well as a mature agile team, a fairly mature and trusting organisation to make this work.

I guess you also need to be working on a product that’s already in production and being updated regularly.