Weak Hypothesis = Weak Startup
You’re probably starting to notice a theme here at the Scientific Startup. We love hypotheses, and we’d even venture to say that the quality of a startup’s hypotheses is a strong indicator of how successful a company will be.
The Scientific Side In our blog on The Scientific Method, we discuss how scientists distinguish between a hypothesis and predictions. In short, a scientific hypothesis is the explanation of a phenomenon.
In a silly example, imagine that your power is unexpectedly turned off. You could hypothesize that the reason for the power outage is that a storm knocked down a power line in your area. It’s a reasonable explanation, and crucially, it is something that can be supported or unsupported through a series of tests.
Observation: My power shut off unexpectedly.
Hypothesis: A storm knocked down a power line in my area, resulting in a loss of power to my home.
After taking an educated guess at the source of the problem (the hypothesis), you can start generating predictions to support your conclusion through testing and experimentation.
Wait for the power company to fix the post and see if the power is restored.
Fix the problem yourself (not recommended) and see if power comes back.
Contact a representative at the power company and verify the hypothesis using their data.
Each prediction can generate a test and data that could support (or falsify) your hypothesis. Every time the test is correct, you can consider the hypothesis supported. Whenever incorrect, you want to keep testing. You can keep running tests to determine if your hypothesis is false or if there are other predictions you haven’t considered. The more success you have with correctly predicting how your hypothesis explains the world around us, the stronger your hypothesis.
When do you stop testing? A hypothesis eventually turns into theory when replicable and empirical data can explain it.
Gravity, for example, is a theory; we don’t need apples to keep falling on our heads to know that it exists. It started, however, as a hypothesis that something is causing items to fall, predictions about the conditions that contribute to the falling items, and decades of successful experiments before the hypothesis graduated to the theory level.
The Startup Spin
You’re working at a startup, not a lab. Most business courses don’t focus on the relationship between theories, hypotheses, and predictions; however, founders are often asked about their assumptions and findings. Let’s break down how these terms are typically used in business and how we can strengthen them by adding a scientific twist.
Starting with the broadest category, an assumption in business is similar to a scientific theory. It’s representative of the “truth” regarding a company and requires a copious amount of evidence to be true.
While established businesses can generate assumptions over time, we frequently see founders building from their assumptions. Investors often hear pitches beginning with assumptions in the first minute of a pitch. “We are going to change the world,” “We are going to be the company that disrupts this space.” or “We will be the Uber of the software industry.”
Assumption: My product is the most efficient B2B accounting platform in the United States.
While these are powerful statements, the adage that assumptions make an a@% out of you and me holds for early companies. Leading with an assumption with no data to support it only shows that you haven’t done your due diligence as a founder. Your investors and stakeholders need more — they need data, and data comes through testing your hypothesis.
The business hypotheses that we usually see are nothing more than predictions, which sets many people up for failure. Let’s dig deeper.
Business Hypothesis: Our product computes taxes 10% faster than the leading competitor.
Why is this an issue? It is incredibly narrow and easily disproven. For example, a competitor could come out tomorrow with something that operates 0.5% faster than you, and BAM, your credibility disappears. By pitching the core value of your product as the result of a single prediction, you’ll spend more time playing defense than offense with your investors and customers.
Let’s revamp this to be a bit more scientific. Here’s a new hypothesis that can be broken down into several tests and provide a variety of proof points for your investors.
Scientific Startup Hypothesis: Customers use our product to save time during tax season.
By creating a hypothesis that several testable predictions can explain, you are far more likely to get the proof points you need to start working towards generating viable assumptions.
Some possible predictions:
Our product computes taxes 10% faster than the leading competitor.
Our UX interface is easier to understand than product X.
Our educational content and preparation give the customers a head start on preparing their taxes.
…. And the list goes on and on.
Taking Action It’s time to look at the type of hypotheses you have used at your startup. Here are a few questions to ask yourself that you can use to turn your assumptions into testable hypotheses.
Have I proven this already? Could I do it again?
Do I have enough data to support my claim?
Is this too narrow? Could it be easily disproven?
Are there other predictions that could explain my hypothesis?
Your investors would rather see concrete proof points at a smaller scale than assumptions generated by passion, rather than science. Building from hypotheses instead of assumptions
Join our Newsletter for bi-weekly content updates. Daily tips can be found on LinkedIn