Tenet of Innovation #2: Fail often; fail early

Innovation is discovery, in both our market and technology understanding. Failing often implies willingness and tolerance for this sometimes-painful discovery process. And we have talked about this to some degree in Tenet #1: Plan on failure; be delighted with success. But although we may expect failure, we wish to get to knowledge quickly, with rapid iterations, so we can make better decisions going forward. The more rapid our learning cycle is, the faster we can refine our approach and hit the sweet spot of the market needs and the technology capabilities. Some people call this “getting to ‘no’ quickly.” In fact, a rapid learning cycle has become a basic principle in many philosophies of product development, such as Agile (Reference 1), Lean (Reference 2), and Flow (Reference 3).

The philosophy of rapid iteration is useless unless we are iterating on the right things. The highest-risk items need to be identified, pulled up early in the process, partitioned, and exercised. The partitioning needs to be done so that the risk is within the individual blocks and not in the interfaces between these blocks. Rapid iteration on these high-risk blocks allows for a higher level of innovation, because the answers to high-risk ideas can be arrived at more quickly and therefore at lower cost. It should then be supplemented by frequent integration with other technology blocks to ensure that the risk is not in the interface. (Note that I have spoken very generically. The same issues are relevant whether the topic is electronic hardware or software or pharmaceutical processing.) Through this process, the product-development team can keep the information “batch” size small, iterate many more times in the same timeframe in order to refine the design and fully understand the risk, and come to a more complete innovative solution.

Although rapid learning cycles feel intuitively correct, we are still often faced with the difficulty of financially justifying this “fail often, fail early” philosophy. To many sponsors, this “plan for failure” attitude is heresy. We need to help with good financial tools to understand the level of failure to be expected and which is financially viable for our industries.

In addition, in order to better value risky projects, some in product-development circles have begun to apply the science of “option analysis” to assessment and decision-making in product-development portfolios (Reference 4). These are being called “real options.” One can think of the process much like the financial vehicles called options. In other words, the act of investing in an early product-development effort allows the sponsor to “buy a call option,” one that would allow for increased investment in the program should the information from the phase warrant further investment. The presence of this “real option” adds value to the product-development effort beyond what would be calculated from the project with standard methods such as NPV (net present value) of cash flows.

Although we can view frequent failure as an illuminating and refining process, we would like the most significant decisions to occur when the amount invested is low. Pulling up the greatest risks to the early stages of a project will allow for rapid, and cheap, iterations. In order to create an environment of tolerance of failure and to produce a business that can afford a high degree of innovation, reduce the cost of failure through rapid learning loops, and encourage early rather than delayed risk-taking.

Years ago, I was involved in the development of ultrasound transducers for medical imaging. One of the areas of great interest was a better transducer match to the human body. I had an idea for a graded impedance piezoelectric material, one that would be made mainly of epoxy at the human interface but would have higher amounts of piezoelectric material in the transducer the further away from the body it was. It sounded as though it would solve many problems. We needed to find out quickly if this was a silly idea or had merit. Through a series of rapid iterations in modeling and experimentation, it became clear that although the acoustics seemed ideal, the electrical properties of the transducer (low capacitance) did not allow for adequate transfer of electrical energy into the transducer. Rapid iteration capabilities in our organization allowed us to test this and other innovative ideas, and in this case, brought us to “no” more quickly.

Failing often (to push forward your understanding) and failing early (to do it cheaply) are essential to the creation of an innovative environment.

1. See, for example, Scrum Primer, by Pete Deemer, Gabrielle Benefield, Craig Larman and Bas Vodde, available through the Scrum Training Institute.
2. See, for example, The Toyota Product Development System: Integrating People, Process and Technology, by James M. Morgan and Jeffrey K. Liker, Productivity Press, 2006.
3. See The Principles of Product Development Flow: Second Generation Lean Product Development, by Donald Reinertsen, Redondo Beach, CA: Celeritas Publishing, 2009.
4. See, for example, “Hybrid real options valuation of risky product development projects” by James E. Neely III and Richard de Neufville, International Journal of Technology, Policy and Management, Volume 1, Number 1, 18 August 2003. See also Options: Theory, Strategy, and Applications by Peter Ritchken for more background.