Keep evidence connected to the commitment.

A stage should indicate what is known about the buyer, not what the salesperson hopes will happen.

Separate confirmed facts, reasonable hypotheses and unanswered questions.

Create another pipeline only when the decisions and stage evidence genuinely differ.

Keep each possible purchase distinct while retaining the shared relationship context.

Use probability-weighted values as a planning scenario, not a promise of revenue or a cash-flow statement.

Change an expected date because evidence changed, and preserve the reason for the move.

Identify the missing decision, decision maker or timing evidence before creating another follow-up.

Separate an observed reason from a convenient assumption and review patterns only after checking record quality.

End with a small number of accountable decisions, not a tour through every deal.

A repeatable task sequence and a triggered automated action solve different problems.

Demonstrate trigger, recipient, action and stop conditions with fictional records before involving customers.

A reply, changed stage or owner decision should trigger a reviewed stop path rather than a blind sequence.

Carry the agreed scope, owner and first delivery action into the project; won does not mean delivered.

Use the commercial record for the buying decision and the delivery record for the accepted commitment.