Split AI Answers Into Facts, Assumptions, and Recommendations

AI writes one confident answer that mixes facts, guesses, and advice. Split it into three buckets: facts, assumptions, and recommendations. The split turns a confident answer into a review checklist.

Facts can be checked against a source: a report, dataset, contract, or product page. Ask where each one comes from.

Assumptions are what AI guessed from context: the cause, the audience, the priority. Treat them as review points, not mistakes.

Recommendations are suggested next steps. Clear writing does not make them verified. They should rest on checked facts.

AI says: This campaign failed because the audience was wrong.

Fact: CTR dropped.

Assumption: the audience was wrong.

Recommendation: test a new segment.

Check first: spend, offer, landing page, tracking, timing, and whether the drop hit every segment or just one.

You may still test a new segment, but only after ruling out the easier fixes.

Use this: Separate this into facts I can verify, assumptions you are making, and recommendations. Keep it short. Tell me what I should check before acting.

It works for summaries, campaign analysis, competitor notes, and reporting conclusions. When you want the weak parts faster, pair it with Critical Review & Harsh Truth.