Use when deciding whether something is worth building, what to build first, what to deliberately not build (non-goals), or whether what already shipped is working — by structuring evidence gathered from real users and real usage, never by asserting it.
