This document discusses how the cost and impact of experimentation depends on technology, business model, and other factors. It argues that experimentation is easiest for web/consumer products using a free business model, but gets increasingly difficult and risky for native apps, enterprise software, and situations where customers rely heavily on stability and predictability. The best practices suggested include obtaining direct customer feedback, using beta programs to test new features or fixes before full release, and leveraging free consumer versions of products to enable more experimentation.