Prior to my last post on music as an experience, I was in the middle of a series of posts exploring business model innovation for the enterprise. A recent exchange on a LinkedIn discussion group has inspired me to pick that back up. In particular, I’d like to explore how a company should go about setting up a minimum viable business to continue the learning process that begins, really, with prototyping.
(Authors note: These business model innovation “articles” are rather long and involved because I myself am trying to work through all the complexity. More like poorly edited papers than blog entries. Proceed at your own risk.)
One of the defining characteristics of a start-up is the ratio of assumption-to-knowledge under which it operates. The goal over time is to invert that ratio. Once that happens, a company can feel confident in scaling the new business quickly and in proportion to the opportunity, then optimizing the business operations to secure the competitive advantage over time.
To get to that inflection point, the analog of R&D investing offers a lot. In referring to R&D, it’s primarily scientific research and the scientific method I have in mind but there are other lessons to be learned as well.
The power and elegance of the scientific method is in its recursive empirical investigation that always allows for falsifiability. A theory is never proven, and similarly, in business assumptions should be regularly revisited and always questioned with a healthy amount of skepticism.
Assumptions, Conjectures, Predictions and Hypotheses
Previously I talked about how business model prototypes can help in the identification and articulation of underlying assumptions (as well as stress testing and scenario analysis to find the right risk/reward profile). The next step is to turn those assumptions into hypotheses, just as you would in scientific research, forming a conjecture and making a prediction to be tested.
One broad assumption can translate into several more specific hypotheses, which is why this translation step is so important. A well-formed hypothesis will be more precise, specific, and easier to interpret.
Characteristics of a good hypothesis include an independent variable that can be manipulated in an experiment and a dependent variable on which its effect can be observed. The hypothesis should also indicate the nature of the relationship (more x leads to more y, for example).
For example, I write this blog both to develop my own perspectives and to create a living record of my thinking accessible to others interested in understanding me, my thinking, and my perspectives a little better. One of the underlying assumptions is that there are actually people out there interested in me, my thinking or my perspective. The hypothesis is that if I make my blog known to people I think would be interested (say recruiters or collaborators on an open innovation project), those people will read my blog.
(For those better versed in the scientific method, please be gentle in correcting any errors I may have made in constructing my hypothesis example).
Independent variable “x” = making my blog known to people
Dependent variable “y” = people reading my blog
Relationship “f(x)=y” = more of x leads to more of y
Interpreting the Results
With my hypothesis in hand, I can begin to think through what are the values for x and y that would support my hypothesis and what are the values that would refute it or be grounds for rejecting it. (Remember, no values will ever prove it, only disprove it.) These values become like leading key performance indicator (KPI’s).
If the evidence I am collecting tells me I should reject a critical hypothesis, that an important assumption is wrong, then I know I need to change something in my business model to account for that new information or my business will fail.
One thing that is important to look for after collecting experiment results is other possible interpretations. Could the increase in y be explained away by web surfers led astray with no interest in my blog? A well designed experiment will minimize the risk of this kind of misinterpretation, but business is not a controlled environment like a lab. Instead, we just have to come up with more experiments to test the alternate interpretations.
Hypotheses and the Minimum Viable Business
Hypotheses provide the design rules or imperatives for setting up a minimum viable business (MVB). Going through the full list of my assumptions-turned-hypotheses, I can identify the most critical ones – the deal breakers and game changers – and design real world experiments – projects – for testing them.
Project can then be grouped logically into programs. The mandate of the minimum viable business then becomes running those programs. This is, in fact, not dissimilar to how a biotech or pharma company might organize R&D to fill its drug pipeline, with programs designed to address specific medical conditions (customer pain points anyone?).
MVB and Insulation
R&D programs are also set up and run separately from the other revenue generating business operations. So too should it be with a minimum viable business.
Allocating a tranche of funding for internal venture investing and then setting up MVB’s as distinct operating units both insulates the new ventures from the interference of hostile vested interests in the status quo (what I call organizational inertia) and protects the core business from the downside risks, both financial and reputational.
A highly effective approach, similar to the one adopted by DARPA, is to set funds aside for research but require business units to fund the final development. For business model innovation, that would mean setting up the initial MVB’s but requiring a business unit to own the P&L for scaling it. If no one is ready to assume those costs, then the venture can be put out “for sale,” by spinning it off.
But wait! There’s more! How R&D performance is measured is also instructive. It is frequently very difficult to measure the return on investment (ROI) of R&D, and consequently, it is commonly treated just as a cost line item. But there is still a return on R&D, and there is a return on experimenting with new ventures.
The problem is that looking at the financial returns of individual investments does not give a good indicator of actual performance. You have to look at the financial return on a portfolio of investments when innovating.
An investment that is a complete financial failure (meaning it generates no direct revenues) may still generate important learnings that contribute to the success of other, later investments. To preserve that connection, financial returns should be measured on the return on a portfolio of investments. All that matters for individual investments is the amount of validated learning it generates, what Eric Ries calls innovation accounting.
A portfolio view also helps in risk management. Hedge funds, R&D organizations, venture capitalists all use portfolio strategies to manage risk.
For business model innovation, the idea is to balance short term and long term investments as well as small payoffs and big bets. By big bet, I don’t mean costly but rather a riskier, high payout. The lean principles still apply, but companies cannot afford to only incubate ideas that may pay off sometime in the distant future. Conversely, game changers will take time and shouldn’t simply be ceded to the competition.
Really the parallels to R&D should be no surprise. Innovation is the new ideas or combinations of ideas that create value in the world. Scientific inquiry serves the same purpose, doesn’t it?