Experiment types in product development, II

Last year I wrote a post discussing some of the common ways to test assumptions and features to make sure that you're only building what's worth putting a lot of effort into.

However, I left off the most important thing to do first. It's the simplest in principle and time involved, but requires more thinking than simply running an experiment.

You must think through how your user will actually use your product, given the change.

That's it.

It's surprisingly common for people to assume they know nothing about how a product will be used. This failure mode is more common than simply assuming that everyone would use the product the same way that you would. A competent product manager understands their users, both through doing user research and applying what they know about their users in their market, and observing how people use other products. They can intuit how the product will be used. They won't always be right, but they will be right an awful lot.

You should have a model in your head of your product and the users of your product. This model will improve over time and the best product managers will have the best models of user behavior in their head.