Prompt engineering guide
with 50 examples

1. State the task clearly — what you want, in one sentence.
2. Give context — background, audience, purpose, constraints.
3. Provide material — paste the document, data or notes to work from.
4. Show an example of good output (few-shot prompting).
5. Specify the format — length, structure, tone, language.
6. Break big tasks into steps — or ask the model to plan first.
7. Iterate — tell it what to change and why.

Writing and communication (1–10)

1. "Rewrite this email to be clear and friendly in under 120 words. Keep the deadline and the ask."
2. "Turn these meeting notes into decisions, owners and due dates in a table."
3. "Draft three subject lines for this newsletter, each under 50 characters, for busy founders."
4. "Explain this policy to a new employee in plain language with three examples."
5. "Edit this paragraph for clarity; show changes and explain the two most important ones."
6. "Write a polite reminder to a client about an overdue invoice of ₹45,000, firm but warm."
7. "Summarise this 20-page report in five bullet points for a CEO, then three risks."
8. "Translate this message into Hindi for a WhatsApp audience; keep it conversational."
9. "Give me three ways to say no to this request while keeping the relationship strong."
10. "Turn this outline into a 600-word LinkedIn post with a strong first line."

Analysis and research (11–20)

11. "From this CSV, find the top three reasons for refunds and show counts."
12. "Compare these two vendor proposals in a table: cost, timeline, risks, gaps."
13. "List assumptions in this business plan and rate how risky each is."
14. "What questions should I ask before deciding between these options? Group them."
15. "Explain this chart to a non-technical audience in four sentences."
16. "Identify contradictions between these two documents and quote the lines."
17. "Create a pros-and-cons analysis of launching in Pune first versus Bengaluru."
18. "Suggest five hypotheses for why sign-ups dropped last week, and how to test each."
19. "Summarise customer reviews into themes with example quotes and frequency."
20. "Answer only from the attached document; if the answer isn't there, say so."

Product management (21–28)

21. "Write a one-page PRD for this feature: problem, goals, non-goals, metrics, risks."
22. "Generate user stories with acceptance criteria for this flow."
23. "Score these ten ideas with RICE; show your assumptions."
24. "Draft ten interview questions about how people last solved this problem."
25. "Turn this interview transcript into insights, quotes and opportunities."
26. "Suggest guardrail metrics for an experiment that increases notifications."
27. "Write release notes for these changes for end users, not engineers."
28. "Play a sceptical CFO and challenge this roadmap; list your top five objections."

Design and UX (29–34)

29. "Critique this onboarding flow against Hick's law and the peak–end rule."
30. "Write microcopy for an empty state, an error and a success message for this screen."
31. "Suggest three alternative layouts for this settings page, with trade-offs."
32. "List accessibility issues in this screen description against WCAG 2.2 AA."
33. "Create a usability test script with five tasks for this prototype."
34. "Write alt text for these five product images."

Coding (35–42)

35. "Explain what this function does, line by line, for a junior developer."
36. "Write unit tests for this function covering edge cases; list the cases first."
37. "Find bugs and security issues in this code; rank by severity."
38. "Refactor this component for readability without changing behaviour."
39. "Write a SQL query for monthly active users by city from this schema."
40. "Convert this API response into a TypeScript type."
41. "Suggest how to make this screen work offline; outline the approach first."
42. "Write a regex for Indian mobile numbers and explain each part."

Marketing and growth (43–47)

43. "Write five app store screenshot captions for this app, each under six words."
44. "Draft a referral message users can share on WhatsApp, under 40 words."
45. "Create a content calendar of 12 blog topics for this audience, with search intent."
46. "Write three ad variants for this offer for first-time investors; avoid hype."
47. "Rewrite this landing page headline five ways for different motivations."

Learning and thinking (48–50)

48. "Teach me this concept with an analogy, then quiz me with three questions."
49. "Make a 30-day learning plan for SQL for a product manager, 30 minutes a day."
50. "Steelman the opposite view of my argument, then tell me where I'm weakest."

• Provide source material and ask the model to stay within it
• Ask for uncertainty: "If you're not sure, say so"
• Request citations or quotes for factual claims — then check them
• Review numbers and calculations independently
• Keep confidential and personal data out unless your tool and policy allow it

Choosing a tool? See ChatGPT vs Claude vs Gemini. For product work, see AI tools for product managers.

Prompt — instructions and context given to an AI. Few-shot prompting — including examples in the prompt. Context — background information the model needs. Hallucination — a confident but false output. Template — a reusable prompt structure.