The purpose of a business experiment is to decide whether a business idea is worth pursuing – either in its original form or a modified form that results from your learning. Clear-headed analysis is called for.
A successful business experiment is a controlled test of one or more assumptions or hypotheses about the business idea being proposed, so as to minimize risks and reduce uncertainty. It serves as a strategy for corporate investment capital to be deployed over large numbers of smaller bets to optimize a portfolio of opportunities.
Before going through the “how-to” of the experiment design, a few points on what not to do:
· Don’t structure experiments to prove your idea is a good one: design for how to get a good read on what would actually happen in roll-out conditions. You may be personally invested in the idea if you originated it, or perceive a personal career risk if the idea is not successful, but you will win over time if you stay focused on getting to a concept that is the best for your organization.
· Don’t focus activities on “winning” but on learning: most ideas go through many iterations before a successful formula results. For example, Redbox movie rentals started initially as a concept to design a self-service convenience store stocked with the most popular items typically sold in retail outlets. From there, the concept evolved several times, modifying product, payment, locations, and other factors
· Don’t engage in black and white thinking: disproving one hypothesis doesn’t mean the whole business concept is flawed. Strategic patience is called for; making the call too fast can actually cost more in the end.
So how to design a good experiment?
Step 1. Identify critical assumptions that should be validated or disproven. Critical assumptions are those which will have the biggest impact if they are wrong and carry the highest level of uncertainty.
Step 2. Develop hypotheses. For each critical assumption, identify what you would expect to happen with your business idea if those assumptions are correct. What are the implications if your hypotheses are proved or disproved?
Step 3. Select metrics. Even if the point of an experiment is to learn rather than to achieve super-fast commercial success, you still need to keep your eye on the ball where traditional business metrics are concerned. So yes, you need learning metrics like time to reach key go/no-go decisions, investment required to reach milestones, and estimates of market demand, but you also need to look at sales, revenue, quality, risk, competition and other market factors.
Step 4. Plan the experiment. Look for the cheapest and fastest way to validate or disprove your hypotheses. If possible, isolate the test scenario from a control scenario. Rely on small, self-contained teams that are not subject to the prioritization of the on-going business. And most important, although you’ll be working on short timeframes and limited budgets, you aren’t excused from the requirement to stick to your plan until you reach an agreed upon stage gate.
In another post, we’ll discuss specifics of these four steps and way