Scientific Method Applied to Products… Loosely.

We’re building a model of how people might behave given a new product offered to them. Will they be thrilled, buy it, use it, tell their friends about it? Will they shrug?

How can we know?

Well, we can’t. At least, we don’t really know until we actually build the product and we measure how well it’s working out. ≈90% of startups fail, mostly due to building something that no one really (most say “wants” here, but that’s not quite right) is willing to pay for and use, mostly because they don’t think it’s worth their effort, time, and money.

A model is first built on assumptions. Assumptions are guesses we make about things upon which our models depend. We can say things like:

• Assuming that gamers need

You can have multiple possible models.

Assumptions need to be validated to determine whether we have the right model.

We ask:

1. Do gamers actually need …

Then the model embodies a hypothesis about such behavior.

Hypotheses must be tested.

To test our hypotheses, we conduct experiments.

If we have tests that are showing the behavior we need for our business and we have assumptions that are valid (that is, they reflect the state of the real world), then we have a model we can action on.

This doesn’t mean stop updating the model. We should continue to make adjustments, retest the hypothesis, make new assumptions and validate them, etc., as we go, to update the state of our model. After all, the world changes continually.