Library › Data AI › Product managers
Email segment-level test results without overclaiming
- For Product managers
- You practice Getting the tone right
- Time about 5 minutes
- Level Intermediate
About this exercise
Your design lead, Kofi Mensah, wants to know how the new onboarding screen did. Write a plain-text email under 100 words with a Subject: line and three short labeled parts: What we saw:, What we can't say yet: and Next:. Report each segment's two rates, call a result real only where the notes say it is reliable, and don't promise the result will hold.
Common mistakes when you ask AI for this
- Treating every lift as real. The AI says both devices improved. Try: Add: "Mobile is a real gain; desktop 6.0% to 6.1% is no real difference."
- Using stats terms. The AI mentions p-values and confidence levels to a designer. Try: Add: "No stats terms."
- Overpromising. The AI says the new screen will lift signups by 20%. Try: Add: "We can't promise the mobile gain will hold."
- Leaving the format to chance. The AI picks its own length and layout unless told the exact shape. Try: Add: "Plain text, no bold or headings."
- Trusting the AI to count words. AI is bad at counting. Ask for a total word limit and it often lands over it. Try: Cap each part instead, for example "each line under 15 words."
What a strong prompt includes
- Names the reader
- Gives the three labels
- Gives the segment rates
- Rules out jargon
- Gives the plan
Related exercises
- Acknowledge a director's critical feedback without defensiveness (Leadership AI · Executives)
- Acknowledge an employee's concern without promising outcomes (People & HR AI · HR and recruiting)
- Acknowledge an executive escalation within the hour (Customer Support & Success AI · Executives)
- Announce a $29 to $39 price increase to loyal customers (Marketing AI · Marketing and content)
- Announce a CFO departure in three plain sentences (Leadership AI · Executives)