Lesson Content

The vague feedback problem.

"Review this and tell me if it's good" produces: surface-level praise, a few minor suggestions, and no deep critique. ChatGPT does not know what "good" means for your specific work product, audience, and purpose unless you define it.

The criteria-first approach.

"I am going to ask you to review [work product]. Before you read it, here are the criteria it should meet: [list your criteria]. For each criterion, rate the work 1-5 and provide specific evidence from this document for your rating. Focus on where it falls short – I want improvement guidance, not reassurance."

Building criteria from purpose.

For a business proposal:

• Does it clearly state what we are asking for and why?
• Is the financial case specific and credible?
• Does it address the likely objections of this specific audience?

For a senior leadership report:

• Is the recommendation clear on the first page?
• Is supporting evidence sufficient without being exhaustive?
• Is every claim within scope of what we actually know?

Role-specific review perspectives.

"Read this as [specific role – e.g., 'a skeptical CFO reviewing a capital request']. What questions would you have that are not answered? What would you push back on? What is missing that you would expect to see?"

When ChatGPT review is not sufficient.

ChatGPT cannot apply: domain-specific expertise you have not provided, knowledge of organizational politics, legal or regulatory compliance judgment, or professional judgment requiring licensure. For high-stakes outputs in these areas, ChatGPT's review is a first pass – human expert review is required before finalization.

ChatGPT file upload for direct review.

ChatGPT Plus users can upload documents directly for review. The same criteria-first principle applies – define your criteria before uploading.