Write a Product Requirements Document from Your Raw Feature Notes

Turns pasted feature notes into a PRD where each requirement has numbered, testable acceptance criteria and missing details become open questions.

You getSummary, goals and success metrics and users and main use cases

Prompt

Prompt
Act as a senior product manager who writes PRDs that engineers and QA can build and test from.

I will paste my raw feature notes right after this prompt. Turn them into a product requirements document.

Rules:
1. Use only what is in my notes. If a detail is missing or unclear (limits, numbers, user roles, edge cases, metrics), do not guess. Add it to Open Questions.
2. Number every requirement (R1, R2, R3). Give each one 2-4 acceptance criteria numbered under it (R1.1, R1.2) in Given/When/Then form, each with a clear pass/fail result a tester could check.
3. One behavior per requirement. Start each with "The system must" or "A user can".
4. Mark each requirement Must, Should or Could.
5. Anything my notes exclude or defer goes under Out of Scope.

Output format:
1. Summary (max 3 sentences: problem, who it is for, expected outcome)
2. Goals and success metrics (only those in my notes)
3. Users and main use cases
4. Requirements: ID, requirement, priority, then numbered acceptance criteria
5. Out of scope
6. Open questions (numbered, each tagged Product, Design, Engineering or Legal)
7. Dependencies and risks

End with one line naming the biggest gap in my notes that would block engineering from starting.

Example output

At a glanceSaved carts PRD at a glanceFreshcart example: requirements, criteria and open questions
Requirements
3
Two Must, one Should
Acceptance criteria
7
Numbered Given/When/Then tests
Open questions
5
Owned by Product, Design, Legal
Repeat-order rate target
34% to 40%
Within 3 months

Example for: notes about a "saved carts" feature for a grocery delivery app called Freshcart.

1. Summary

Regular shoppers rebuild the same weekly order by hand. Saved carts let logged-in customers store a cart and reload it in one tap. The goal is faster repeat orders.

2. Goals and Success Metrics

  • Raise repeat-order rate from 34% to 40% within 3 months (from notes)
  • Cut median checkout time for repeat orders (no target given, see Q2)

3. Users and Main Use Cases

  • Weekly shopper: saves a standard order and reloads it each week
  • Household planner: keeps separate carts, e.g. "Weekday" and "Party"

4. Requirements

R1. A user can save the current cart with a name. Priority: Must

  • R1.1 Given a logged-in user with items in cart, when they tap Save Cart and enter a name, then the cart appears in Saved Carts.
  • R1.2 Given a guest user, when they tap Save Cart, then they are prompted to log in and no cart is saved.
  • R1.3 Given an empty name field, when they tap Save, then the button stays disabled.

R2. A user can load a saved cart into the active cart. Priority: Must

  • R2.1 Given a saved cart, when the user taps Load, then all in-stock items are added at current prices.
  • R2.2 Given an item now out of stock, when the cart loads, then that item is skipped and listed in a notice.

R3. A user can delete a saved cart. Priority: Should

  • R3.1 Given a saved cart, when the user taps Delete and confirms, then it no longer appears in Saved Carts.
  • R3.2 Given the user cancels the confirmation, then the cart remains.

5. Out of Scope

  • Sharing carts with other accounts (deferred to Q3 per notes)
  • Scheduled automatic reorders

6. Open Questions

  1. What is the maximum number of saved carts per user? (Product)
  2. What checkout-time target counts as success? (Product)
  3. When loading into a non-empty cart, merge or replace? (Design)
  4. Should loading suggest substitutes for out-of-stock items? (Product)
  5. How long are saved carts kept for inactive accounts? (Legal)

7. Dependencies and Risks

  • Depends on the inventory API returning live stock per store
  • Risk: price changes since saving may surprise users at checkout

Biggest gap: merge-vs-replace behavior (Q3) is undefined and blocks the core load flow.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Given/When/Then criteria with pass/fail results give QA something to test instead of vague intentions.
  • Banning guesses and routing gaps to tagged open questions keeps invented details out of the spec.
  • The closing blocker line tells you which question to resolve first before engineering starts.

Use when you have messy meeting notes, Slack threads or a feature brief and need a buildable PRD. Paste the notes right after the prompt in ChatGPT, Claude or Gemini.

Common edits

  • Ask for the requirements as a Jira-ready list with one story per requirement.
  • Add a section for non-functional requirements such as performance or accessibility.

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Product Management