Promises are often built on mixed evidence
Some inputs are facts, some are estimates, some are historical averages and some are optimistic assertions. A good decision distinguishes them.
Evidence quality is contextual
A delivery date may need a resource plan and dependency confirmation. A security commitment may need architecture evidence. A margin assumption may need a cost model. The right evidence depends on what is being promised.
Unknown is a valid state
The pressure to close a deal can turn missing information into assumed information. Marking an assumption as unresolved is more useful than pretending the evidence exists.
Confidence should be traceable
If a leader asks “why do we believe this?”, the answer should point to evidence, owner and date—not to a general sense that the team is comfortable.
Evidence improves repair
When the problem is weak evidence rather than a bad term, the smallest repair may simply be to obtain the missing proof before the promise is made.