The reps who get the most from AI treat the first output as a first draft that needs editing — not a finished product to accept or reject. This mental model changes how you interact with the tool entirely. Instead of evaluating whether the output is "good enough," you're identifying specifically what to improve in the next iteration.
The most effective workflow for most sales tasks:
Pass 1: Run your structured prompt. Get the initial output. Read it once and identify the 1-3 most important things to improve. Not a comprehensive critique — just the most important things.
Pass 2: Give specific refinement instructions. Not "make it better" or "make it sound more human" — those are vague and produce vague improvements. Instead: "The opening line is too generic. Rewrite it to reference the prospect's specific LinkedIn post about their outbound team growth. Keep everything else."
Pass 2 output is almost always significantly better than pass 1. This is where reps who iterate well pull ahead of reps who accept first drafts.
For cold email: Is the opening line specific or generic? Is the value statement in the prospect's language or yours? Is the CTA frictionless?
For analysis outputs: Are the risk flags real or generic? Do the recommended questions match the specific gaps identified?
For LinkedIn posts: Does the first line stop the scroll? Is there at least one specific detail that makes this post feel real rather than manufactured?