Recently I had the opportunity to take Launching Technology Ventures with Tom Eisenmann at the Harvard Business School, and learn the latest and greatest about the process for building capital efficient startups using the “lean methodology.” Lean is a movement started by Eric Ries that encourages entrepreneurs to design cheap experiments to test their products and positioning in the marketplace and get real data, rather than relying on the founder’s grand vision which can often lead to building a business that is 10, 90, or 180 degrees off from actual market needs.
My project for the course involved a “science-based” hardware business. I worked with a university professor in Switzerland to spin a fundamental technology out if his lab that we believe could revolutionize on-chip and off-chip digital communications. Tools and techniques that lean methodology suggests using include highly agile product development cycles, launching early, building “dummy features” to see if users interact with them, A/B testing, and watching customers actually use your product. As such my initial reaction was that lean had no place in the business I was working on, where our customers will in many case be large semiconductor or hardware manufacturers (not a group that can easily be used in a beta test), and our product development cycles are long (since we’re doing fundamental research in a lab and fabricating something physical vs. writing application code in an Amazon EC2 cloud).
What I learned however, was that even though lean will need to develop a different set of tactical recommendations about how to run a product development or marketing process in such businesses, the underlying framework is still very useful for entrepreneurs in this space:
- Develop a hypothesis for what you believe the right answer will be.
- Design a cheap experiment that serves as a falsifiable test to see if your hypothesis can be disproven.
- If the answer is “it can be disproven,” then reform the hypothesis and re-test it. Otherwise, form a new set of hypotheses that further your understanding.
The key here is appropriately designing the falsifiable test so with some degree of certainty you can challenge your hypothesis, rather than just “gathering data” through market research to give yourself a better hunch. This guidance helped me and my colleagues ask specific questions of market experts such as “is there anything that would make it impossible for 5 engineers and a $4M capital base to develop XYZ product for ABC market in 2 years?” The response we heard was “well before you could start development, you would need cross-licensing IP agreements with either company D, E, or F, and such an agreement typically takes at least 10 months to put in place.”
This helped us understand that we couldn’t simply solve the problem of hitting market at what we believed was the right time by doubling the number of engineers or capital, but that it was a long pole external item. From there we could ask the next set of questions such as, “how would we go about finding the person we need to get on board to get the deal done in 4 or 5 months?” (a new hypothesis that such a person must exist). This iterative process helps us systematically remove risk from the business without spending a lot of money.
I encourage people who are passionate about the lean methodology to develop more specific techniques for deep technology entrepreneurs so that everyone doesn’t have to design their own experiments from scratch. While I have great belief in the future of Internet businesses with strong network effects to both change the world and return well for their investors and operators, I feel it’s undeniable that revolutions in clean technology, biotechnology, and data infrastructure have a critical and unique role to play in the sustained growth of entrepreneurship and United States GDP.
