Why Scaling Agile Doesn't Work

YOUTUBE 2zYxWEZ0gYg Jez Humble presents at GOTO 2015

Humble argues that a lot of organizations create scrum in the software creation stage, but that the real problem is that the feedback loop is so long that they can't derive value from it.

We spend a lot of time on discovery and estimation and fixating on cost, but cost is not as important as (1) will the customer use it, and (2) will the project get cancelled.

"Black Swan Farming using Cost of Delay" a study of the cost of not delivering a requirement ranked per requirement.

On reflection, this is perhaps what I was frustrated by in Storypark as they spend a lot of time fixating on estimation and cost and I felt we didn't spend enough time reflecting on our understanding of the problems we were solving.

Maersk published an amazing study called Black Swan Farming where they ranked different requirements in order of money lost while the feature was not being delivered. Humble recommends this paper as a method of implementing agile at scale.

User's don't know what they want. Requirements generally come from HIPPOs.

How to move forward

Don't optimise for the case where we think we are "right"

Focus on value, not cost. Find out what is highest value and gather information to justify the value of what we're building.

Create feedback loops to validate assumptions in our delivery process. How can we get this feedback as fast as possible so we can adjust our approach.

Make it economic to work in small batches. This is important because we get feedback at high frequency which allows us to course correct much more easily.

Enable an experiment approach to product development. This lets us use the scientific method on process improvement and product development.

Measuring and Alignment

18:46 Impact Mapping, a way of figuring how to work backwards from the outcome.

Don't fixate on the artifact. Create a shared understanding with the team about how it affects things. Get everyone together then pick one.

We believe that [building this feature] [for these people] will achieve [this outcome]. We will know we are successful when we see [this signal from the market]

This is a really nice alternative to the format "As a [Stakeholder], I would like to [Action] so that I can [Outcome] (added for searchability)

Humble took this template from Jeff Gothelf's talk Lean vs Agile vs Design Thinking which has been given a few times.