Guest Post by Robert G. Cooper
Editor’s note: In this excerpt from the September-October issue of the CIMS Innovation Management Report, Robert Cooper, “father” of the Stage-Gate model of new product development and ISBM Distinguished Research Fellow at the Smeal College of Business Administration, Pennsylvania State University, discusses the challenges of creating an Agile-Stage-Gate hybrid approach.
More than 10 years ago, the software world began to integrate Agile into Stage-Gate. Agile, specifically the Scrum version, is a set of software development methodologies that breaks development into a series of short, iterative and incremental sprints, typically one to four weeks long. The gating model provided strategic direction to projects, dealing with the “what” and “why,” while Agile focused on task execution within the stages, namely the “how.” Very recently, a handful of early-adopter manufacturing firms such as Honeywell and LEGO began to use this Agile-Stage-Gate system for physical or hardware projects. Initial results have been dramatic, with significant increases in productivity, reductions in time-to-market, and a much higher likelihood of getting the product right.
But there are also challenges. The “elephant in the room” for many companies is finding the resources. Agile-Stage-Gate requires dedicated project team members who are full time on one project and co-located in one room. The requirements for rapid time-boxed sprints, daily scrum meetings, and delivering something at the end of each short sprint mean that the system simply won’t work when team members are spread across many projects, which is normally the case in physical product development.
Companies have found work-arounds for these challenges. For example, a major Swedish packaging and equipment firm began using Agile-Stage-Gate in one of its divisions four years ago. The company has quickly learned that Agile requires dedicated team members. As the senior project manager explains, “First we tried teams with people dedicated 50 percent of their time, which did not work at all. Then we moved to 70 percent dedication, which was better but not optimal. Now, the rule is a maximum of one other project per person, taking no more than 30 percent of their time. This works, but we might even have to move to 100 percent!”
If companies have no choice but to ignore this “one-project” requirement, they should try to limit the maximum to two projects per person, with each team member devoting the same two days per week to the one Agile project so the team can be together for two weekly scrums. By using longer sprints, as many manufacturers do, two scrums per week may be all that is needed.
Another way to help ensure success is to be more selective about projects. Agile-Stage-Gate is best suited to more dynamic development projects facing fluid markets and changing customer requirements—typically more innovative and ambiguous projects. Consequently, not every development project is a candidate for Agile-Stage-Gate. Most manufacturing firms limit the approach to their most important, larger, more innovative development projects.
The development team starts by agreeing on what they can accomplish in the next time-boxed sprint, and maps out a task plan for the sprint; once the sprint is underway, the dedicated, co-located team members meet daily in “scrums” to make sure all is on course. Product increments are developed in each sprint, regularly demonstrated and validated with customers and with management. With feedback in hand, the team then plans and begins their next sprint. Product requirements and technical solutions, and even the project plan, thus evolve over the development cycle.
For example, at Chamberlain, which makes remote-controlled devices for the home, the Agile-Stage-Gate approach is used only for the larger, major revenue-generating projects, which represent about 20 percent of the projects the company is working on.
It’s also important that all divisions be on board. Non-R&D areas, such as marketing or manufacturing engineering, often find it difficult to commit dedicated effort to the project. Solutions include training, an effective change management effort, and proper project prioritization and creative resource allocation. For example, marketing people at LEGO, who are spread thinly, commit to a dedicated effort for a single two-week sprint–that is, spend all their time on one project, doing a market analysis or voice-of-customer study, for example. When that dedicated sprint is over, they return to their regular jobs, but continue to spend part of their time on the Agile project on a non-dedicated basis.
Fencing-Off Resources and Strategic Buckets
Resources are often not available to do these larger, more innovative projects simply because resources are totally consumed by the many small projects. One solution is an organizational one, namely to fence off an “innovation group” to work 100 percent on major developments or breakthroughs. A second solution is a portfolio approach, namely strategic buckets. In this scenario, management makes a strategic decision to set aside a percentage of resources for different types of projects, including candidate projects for Agile-Stage-Gate. Various categories or project buckets are defined, from maintenance projects to true innovations. Then the business’s executives make strategic decisions about what proportion of their R&D resources goes to each bucket, setting aside “resource buckets” for each major category.
Next, active and proposed projects are categorized by bucket, and ranked-ordered within each bucket until there are no more resources in that bucket. When used over time, the method ensures that resources are reserved for innovative projects; it causes the development portfolio to mirror the strategic priorities of the business, and it protects higher risk, innovative projects, which are not compared to predictable, smaller ones.
Fewer But Better Projects: Gates With Teeth
Most manufacturers have far too many projects, and often too many minor projects, for the resources available. This makes finding dedicated teams for Agile projects almost impossible. The solution: Introduce a decision factory mentality into gate meetings. Gates should be Go/Kill decision points where investment decisions are made, not just milestone reviews. The goal is “gates with teeth”: Kill the weaker projects! More discrimination in terms of projects approved will result in better projects, more focused teams, and shorter time to market.
Managers need to commit the resources at gates. Gates are not just project approval meetings but decisions to commit resources to a project leader and team. When projects are approved but resources are not, the result is a hollow Go decision. And they should keep a tally of resources committed versus those available. Putting a limit on projects consistent with resource availability prevents pipeline gridlock and means that dedicated teams will be available for Agile projects.
While Agile-Stage-Gate does deliver dramatic results, it also requires some changes in how people work. Dedicated teams are a requirement, which means more focus, better project prioritization, and doing fewer but better projects. There is no one magic solution to securing dedicated teams, but several approaches applied concurrently—project limits and selectivity, strategic buckets, and tougher Go/Kill decisions—address the challenge.
Tools To Make Tougher Go/Kill Decisions
NPV (net present value): NPV is theoretically the correct financial method for Go/Kill decisions and does provide the impression of rigor. But reality is much different: The handful of firms that have compared predicted versus achieved NPVs find huge errors in predicted values. Further, ranking project by NPV often yields the wrong priorities. Finally, NPV does not deal with resource constraints, so that any project that “hits the hurdle” is potentially a Go!
The Productivity Index: This index, an extension of NPV, is based on the theory of constraints. To maximize the value of your portfolio when resources are constrained, take what you are trying to maximize—the NPV, for example—and divide by your constrained resource. One example would be the person-days required to complete the project. The result is the productivity index, which gauges the “bang for buck.” Then rank your projects by this index until out of resources. With discipline, the result is both a maximized portfolio value and a limit on the number of projects—the pipeline is not overloaded.
Profiling or scorecards: Although scorecards are not the most popular method, they produce surprisingly good results. A scorecard is constructed using those factors that are known drivers of success. Then use the scorecard at your gate meetings to rate projects. The gatekeepers (not the project team) score the project on six to ten key evaluation criteria. The score, along with the rich discussion as gatekeepers debate their scores, leads to a much-improved decision where many factors are considered, from competitive advantage to technical feasibility.