Packages come in all shapes and sizes!


Buying a package to perform a major function in an organization is rarely as easy as buying and implementing an app on your smartphone.  Package implementations often include:

  1.      Integration with other applications
  2.      Configuration
  3.      External customizations
  4.      Conversion of current data
  5.      Training users
  6.      Communication and change management activities
  7.      Business process re-engineering
  8.      Buying and installing the package (we will wrestle with SaaS and the Cloud later in this series)
  9.      Hardware and network changes

Purchasing Microsoft Word and installing it on your computer might fall not be as complicated as this list suggests; however, installing packages like PeopleSoft, SAP or Workday will require most if not more than this list of activities. Implementing COTS software for anything substantial is a complicated endeavor (albeit less complicated than building the whole thing yourself).   The most common reasons given for why an agile approach to these efforts won’t work are three-fold.

  1. Coordination amongst all of the teams and organizations
  2. Transparency between the organization and COTS vendor(s)
  3. Implementations for COTS are typically are either comprised of a big bang or a few big’ish banglets


COTS implementations for major packages are large projects and involve many different teams.  Many of the same solutions that we noted our recent article on Large Mainframe Projects are germane to this scenario.

  1. Adopt a Scrum of Scrum – SoS’s foster focused cross-team coordination.
  2. Implement visual management using a Kanban board and other information radiators.
    Make sure everyone can see the progress toward the goal.  This will help involvement and will support self-organization.
  3. Follow a common cadence of joint planning at a strategic and tactical level.
    Consider the 90-day planning increment adopted by SAFe combined with everyone participating in team level planning at the same time (every 1 to 3 weeks based on a common sprint/iteration cadence).
  4. Leverage the package architecture to seed the projects architectural runway.
    The COTS package will have an implementation architecture; use that architecture to inform the organization’s overall architectural runway. The subtitle of the movie Dr. Strangelove put is best, “How I Learned to Stop Worrying and Love The Bomb.” When you buy a major package you often have just bought an architecture for a big piece of your business.


Transparency might be the hardest problem to solve.  Each organization will have their own methodology for implementation.  The level of transparency between the organizations often is defined in the contract which spells out how the organizations will interact if you are going to agile to coordinate and communicate it must be in the contract.   We will revisit contracts later in this theme; however, building the many of the coordination steps into the contract before singing will solve much of the transparency problem.  Another recommendation is to review the timing for when the COTS vendor releases draft release notes.  The earlier and more in-depth the draft release notes, the easier it will be for teams to integrate the flow of work into standard iteration planning and execution.


It is difficult to implement a package in production a little bit at a time.  Most organizations “implement” all at once or in large chunks.  An agile approach often cannot impact that trajectory. However, much of the activity (integrations, conversions, configurations, coding and more) can be completed and tested using all of the lean and agile ideas explored in earlier articles, such as Agile Where Agile Fears to Tread: Maintenance, small pieces completed in short iterations that can be judged production ready.

Coordination, transparency, and implementation make implementing and maintaining major COTS packages complicated.  No methodology will make these efforts easy, but methods and techniques that improve communication and generate faster feedback will improve coordination, transparency and reduce the risk of the typical big bang implementation.