A common failure mode for organizations attempting to adopt an Agile style of software development occurs when an attempt is made to “Scale Agile”. Suddenly, the organization finds that there are scheduling problems between teams. Delivery team members suddenly find that they are required to serve on several teams at once. Dependencies surface, and teams find it difficult to come together in a common cadence to produce working software in a continuously delivered fashion. Many times, these issues become so grave that the organization reverts back to the Waterfall model that they came to hate, but at least understood.

This session explores Agile scaling concerns, and places particular emphasis on an architecturally significant distinction in the software to be created, and the components produced to allow the software to be created. That distinction revolves around cross cutting platform concerns versus product feature creation concerns. We will examine the distinctions and explore solutions that should help your organization get past these issues when it comes to portfolio management, by paying attention to extrinsic versus intrinsic value metrics.


Outline/Structure of the Talk

  • Definition of the problem for the session, specifically excluding the standard "scaling Agile" problem, and specifically including the problem related to dependencies between core component silo teams and multiple product development teams [5 minutes]
  • Definition of what are Product Concerns and what are Platform Concerns [5 minutes]
  • Brief explanations of the grading system used (the 7 Deadly Lean Wastes) to compare six solutions to the problem [I've coached organizations where various flavors of each solution were used, and I'm using their results but witholdng identifying information] [10 minutes]
  • Compare and contrast the six solutions [15 minutes - I will cull down this section and get more summary on grading - I planned on reformatting this section anyway, and there is no reason to speak so much on obviously poor things like returning to waterfall for because the organization is finding scaled agile to be hard]
  • In depth explanition of the "Scrum Everywhere with Core Component Kanban Teams" solution and how to make it work [5 minutes]
  • Concluding remarks and final thoughts [5 minutes]

Learning Outcome

  • Use of pure waterfall silos, pure Scrum silos, hybrid Scrum product teams with core component silo waterfall teams, hybrid Scrum product teams with core component silo waterfall teams and "Guest Stars", and even spinning off core component silos into companies all fail organizations.
  • Lean thinking explains the failure modes and helps us keep from repeating those failures
  • A more complicated model involving core component teams running in Kanban and embedding with Scrum product teams works, so long as you also have core component teams using Scrum for development of future core component capabilities
  • These concepts are explained, and experience with how this has worked in enterprises I've coached is explored 

Target Audience

Managers who organize team makeup and are wondering how to engage individuals from key component silos

schedule Submitted 5 years ago

Public Feedback

comment Suggest improvements to the Speaker
  • Naresh Jain
    By Naresh Jain  ~  5 years ago
    reply Reply

    Howard, I really think this is a powerful proposal. Can you please update your proposal with a time-wise breakup of your topics under the Outline section? Also, as pointed out on your other proposal, 90 mins talk is really hard to keep the participants engaged. Also given that there would be multiple tracks, I doubt many people will want to spend 90 mins on one session.

    • Howard Deiner
      By Howard Deiner  ~  5 years ago
      reply Reply

      Hi Naresh!

      I'm comfortable in delivering this in 45 minutes.  When I reflected back on the two times I've given this, I realize that I tried to teach too much on basic Lean principles.  I think that I would improve the session by mentioning the ways I want to compare and constrast the solutions to scaling, but assuming that people already know a bunch about at least what the 7 lean wastes look like in software (the Poppendick stuff).  I also have wanted to cull down that section just in terms of the word density.  Less is better, and it gives me more options to stay on track timewise when there is less on the slides.  Key terms only, with just enough verbal guidance will allow me to breeze through the obviosly bad solutions.  The audience already knows those are bad, so why "beat a dead horse?"

      Thanks for the comments!!