You have built a $100M company from the ground up, taken if from an ad hoc wildcatter to disciplined product development machine. Your phase gate system weeds out bad concepts, lackluster products, and has generally been successful. However, more and more your products are finishing development only to discover the market has shifted and launch performance has underperformed predictions. No matter how much more upfront work you do, nothing improves. By the time you finish development and gain market feedback, 80% of the investment is sunk, and there is no recovery.

Your developers talk you into trying Agile Product Development, and you agree. The first few products come out well and sales are solid. As the team gains confidence and independence it starts to feel out of control. The team begins to fail when it chases markets that are too small and some projects never seem to finish. Your phase gate controls have all broken down and are not managing risk.

What do you do?

Jim Highsmith (Agile Project Management), has a chapter on governance that deals directly with this problem. Jim points out that in the old phase gate systems, governance and operation (product development) are tightly coupled. Solving your problem requires "separation of governance from operations and then loosely coupling them." Governance is a linear process for managing investments and risk, operating much like real options. Agile works on the principle of iteration, experience, and feedback.

The crux of the solution is in redefining the stages and gates to align with Agile. Traditional phase gate has 5 stages: Opportunity Identification, Concept Generation, Concept Validation, Development, Launch. This clashes with Agile which makes less assumptions and depends on exploration and feedback. Highsmith suggest using 4 phases and gates: Concept, Expansion, Extension, Deploy. Each phase mitigates risk. During Concept the core product ideas are worked out and a few critical iterations are completed. During Expansion high value features are built and as much risk as possible is is driven out. Extension completes the product with minimal risk and better foresight into costs and market acceptance. Deployment puts the product into the market.

The core difference is the Agile team is building a deployable product in every iteration all through all the phases. This reduces the time from requirements to feedback so that the process can respond to market dynamics, rather than the traditional phase gate that builds a deployable product at the end after most of the investment is made. At each gate, executives evaluate the project as an option on future investment. Do we make the investment and buy the next option? Do we cancel the project? Do we wait and see?

The alternative phase gate system allows the executive team or portfolio manager to make linear investment decisions while the team runs an iterative process. This combination will also speed up gate decisions. A classic phase gate requires all work to line up for a moment in time to product documents and presentations for a gate meeting. This interruption causes value flow to stop. Agile iterations are time boxed with deployable product at the end of each time box, so the team is already lined up in time. Evaluation is simplified because the team can demonstrate a product to customers and the executive team. So not only does Agile speed up delivery of value and responsiveness, it actually speeds up getting through the gates, as long as the executives get in alignment with the heartbeat of the project.

Much of the traditional phase gate system can be salvaged, so this is not a start over from scratch. Opportunity Identification and Concept Generation can be placed in the new Concept Phase. Much of the Concept Testing can be placed in the Expansion phase, however a deployable product is created each iteration, so Concept Evaluation, Prototyping, and Market Testing all become the same thing. The key is to redefine the phases and gates to complement Agile.