Blackwater Falls in WV

Falls or rapids?

I was recently asked about interfacing an agile team with the rest of a “waterfall” company. This is not an uncommon theme in the questions that I receive.  A synopsis of the scenario I was given at the start of the conversation follows:

I’ve been working at an organization that has an interesting structure. Our part uses Agile while the larger umbrella uses a waterfall approach. The team I lead as a PM/Scrum Master performs well but is bogged down with commitments made to the waterfall side of the house. This means that even though we are a scrum team, we have release dates that were promised at the start of the year that we’re bound to. What advice can you give me to try to overcome this?

There are several threads in this simple scenario that have to addressed to “fix” the problem in the long run (which means this is not simple at all):  

  1. Prioritizing Work Entry 
  2. Postponing Commitment
  3. Engaging Executives On Agile Values
  4. Value Chain Analysis

A long term solution will need to synthesize ideas from each of these threads.

Prioritizing Work Entry:

One more piece of information before I start into a cleaned-up version of my initial response.  The person asking the questions and I have worked together before — we have shared history.

Suggestions:  

  1. You could try to hold your breath until the waterfall approach from the home office goes away. This has a low probability of having a positive impact, but it is dramatic. Do not try this at home (the same thing would be true for hunger strikes)! In order to generate change, you will need a concerted discussion about agile values and behaviors. Facilitating this discussion to generate an action plan will not be quick and which won’t help your team in the short run. 
  2.  A better short-term solution within the span of control of both you and your product owner is to leverage your product backlog as a gate to control the work coming into the team. There are several steps needed to implement this statement:
    1. You need to have a backlog.
    2. You need a product owner.
    3. The team cannot be allowed to accept work that is not prioritized on the product backlog by the product owner. If this is difficult because sprints are too long, shorten the sprint length. Start with a discussion of how to address this item during your next retrospective, your team will have ideas for addressing work that appears outside of planning. 
    4. Work that is committed to by the home office and “placed” on your backlog with release dates need to be identified (use a flag in your ticket/story tool) and prioritized just like any other piece of work.
    5. Add kanban to your scrum. Establish WIP limits to each step in the process and take work from the backlog only when you have the capacity to work on it.  Scrumban is a great adaptation that allows you to establish flow and visually keep track of work.  
    6. Flag the items promised by the home office and use that flag as a prioritization tool.  Flow and flow metrics will allow you to predict when those items can be delivered and when they need to be pulled into your flow (this would impact prioritization).  Throughput accounting rather than story points Use the work entry process to protect the team and to deliver based on the waterfall promises.  Use throughput to gauge how much work you can do (not story points and velocity).  

This feels like a bit of a prescription, however, it is an outline that will need to be customized based on your specific team and stakeholders. In a nutshell, start by establishing control over your work entry process. If you can control how the team accepts work you can satisfy your corporate requirement or at least can have a good conversation about priorities.

This is the first step and will help allow your team to increase their performance to an even higher level.  The next step is to build an understanding of how postponing commitment actually increases throughput and value delivery (we will draw from Daniel Doiron’s response to help frame that idea next).