September 26, 2018 // By Michael Dougherty
Addressing the often overlooked and undervalued effort of rationalizing your portfolio of applications taking new concepts of Agile, Lean and DevOps.
There are many fantastic articles on Application Portfolio Management on the Internet. For instance, "How to Rationalize Your Application Portfolio" in CIO.com by Thor Olavsrud and "How to do Software Rationalization Right" by Chris Doig are a keen starting point for justifying Application Portfolio Rationalization (APR) and the process at a high level.
As Thor puts it, "application bloat" is a significant and growing problem in Enterprises. This is as true now as it ever has been, especially when companies merge and acquire each other. Thor's example with the Alcatel/Lucent merge is an excellent case study. However, actively addressing "application bloat" is often ignored and overlooked by senior executives. Why? It's simply not very exciting and although significant, the benefits often take major research to fully explain and realize.
Since you are reading this blog, my assumption is you recognize the significant cost in licensing, server maintenance, support and operations of outdated, overriding and colliding applications.
Chris provides a comprehensive list of common causes of application sprawl along with typical problems and ownership costs. I'll leave you to read up on the economic pain level managing this application sprawl costs your organization every day.
Modernizing the Portfolio Rationalization Process
Coming from the world of Project, Program and Portfolio Management, I've seen these Portfolio rationalizations start on the best of intent with detailed delivery plans and specification analysis, yet typically run way over budget and extended for years, crushing themselves in their own complexity and lack of clarity. They are rooted in Waterfall delivery at the top level, negatively impacting business and economic benefits. Even with that, the Return on Investment (ROI) and Total Cost of Ownership (TCO) along with other financial measurements still makes it a "net positive" expense overall.
Applying waterfall to an ocean of apps is expensive. Companies feel a significant amount of pain in the current ways of comparing business value, ownership costs and mapping. By applying today's concepts of Agile, DevOps and Lean software delivery, there is a better way to deliver and its in funneling that ocean into smaller pools (i.e. batches). The key approach shift is to reduce the cost of delay at the portfolio, program and project levels while maximizing economic benefit. Note the value of standardizing and reducing variability is listed as "Stage 2" in the 2018 State of DevOps Report and ties directly to managing your Portfolio.
Implementing the Portfolio Rationalization
So how to accelerate Application Portfolio Rationalization (APR) using modern Agile, Lean and DevOps (also known as DevSecOps with Security) practices?
Start with the standard concept of a governing body. This governing body would be smaller in size consisting of no larger than a standard Scrum team in size (maximum of nine people), with preference to the smaller size. Let's call this governing body the Accelerated Portfolio Management or "APM" for short. The APM would consist of Enterprise Application Owners to handle the business side, Enterprise Architects for the technical side and an Enterprise Program Manager (or Agile Project Management Office) to handle the process side. There may be a Product Manager, Director of Technology and other roles based on your Enterprise context, but in this case, "less is more". Invite too many people and the entire decision-making process will slow down your APR efforts and thus increase your overall costs.
Shown successful within the IT industry, especially in maintenance and support, Kanban boards helps visualize all those applications for quick assessment and perhaps more importantly, uses a pull-based system that limits the work in progress for each step progression based on meeting certain criteria. This Application Portfolio Kanban (APK) helps quickly focus on those most economically valuable apps first and leave the others for research later.
To promote smaller batch sizes and retain simplicity, the shown above APK would consist of five primary lanes:
The "entire ocean" of apps start here. Only a minimum amount of research has been completed. Applications that move on to the next lane will be decided by the APM based on strategic themes and current applications identified as the most critical due to compliance, ending of support, high maintenance costs, etc.
The Review lane filters the top applications from the funnel based on set criteria formed by the APM. The focus of the Review lane centers on a quick benefits analysis via an Application Hypothesis Statement with sufficient information for the APM to compare and prioritize applications to be candidates to move on to the next APK lane. The method used combines Weighted Shortest Job First (WSJF) with appropriate categorization. Like all Kanban, a Work In Progress (WIP) limit based on the capacity of the ACE shall be maintained.
The Analyze lane filters top candidates from the Review lane as determined by the APM. The focus of the Analyze lane centers on the costs of rationalizing the applications including a technical complexity evaluation resulting in a refined WSJF under a standard WIP limit. The APM will finalize and formally approve the applications that will be accepted into the next lane.
The Backlog lane provide the list of Applications next approved by the APM to devote delivery effort. This does not mean work starts immediately but is in a “ready state” for a delivery team to pull once their previous work has completed. That includes all potential applications approved to be delivered within a set timeframe agreed by the business. Applications will fall under four categories:
- Invest (new re-write under modern technology)
- Rationalize (update for functioning on modern technology)
- Re-platform and Retire (entirely move the application to modern technology and retire the original)
- Retire and Shutdown (completely remove the application since its use has become outmoded)
To reach “Done”, the original goal and outcome hypothesis is validated for the rationalized application and used to determine the success level. This is then measured and used for learning improvement ideas for that application and full APR process itself, feedback back into improvement actions for the APM and APK process.
The Application Rationalization Portfolio process is complicated, costly and very time consuming. Enterprises often do not give APR the attention deserved since the outcome can provide high benefits and productivity while meeting typical expectations of decreasing overall portfolio costs. Often, enterprises are challenged by the sheer size and complexity of rationalizing a portfolio, especially for larger Fortune 500 corporations, so streamlining the analysis overhead by visualizing and abstracting at the right amount will accelerate results and therefore further support.
Don’t ever think APR is a “one and done”. This is a continuous effort recommended for all major enterprises!