Itamar Gilad

Never assume your competitor knows what they're doing. 'They have this feature so it must be a good idea' is one of the worst forms of validation. It almost never works.

The confidence meter: self-conviction = 0.1/10 confidence. A competitor having the feature = low confidence. Actual user tests and experiments = medium to high confidence. Most teams stop way too early and call it validated.

Google+ had ~1,000 people at its peak — the size of Android. It was run like a secretive Jobs-style project. And it produced zero impact on Google's revenue or Facebook's growth. That's what opinion-based development at scale looks like.