When you’re not selling a physical product, you’ve got three ways to monetize.

To prove causation when you can’t run an actual experiment, introduce pseudo-randomness. In particular, instrumental variables can be used to mimic experiments and isolate causal effects to help reveal causation.

The “cold start problem” evolves into ever-shifting imbalances in supply and demand.

To assess the risk of zero, use these four crucial questions.

You shouldn’t blindly follow every decision it makes. You must combine data with logic and business sense to make the best decisions.