Skip to main content
Product Manager Prompts

ChatGPT Prompts for Product Managers

Write PRDs faster, structure user stories properly, prioritize your roadmap with frameworks, and prep for stakeholder reviews — all with AI prompts built for real PM workflows.

12 prompts|By PlugMonkey Team, Editorial

Product managers write constantly — PRDs, user stories, OKRs, roadmap justifications, sprint reviews, stakeholder updates, competitive briefs — and most of that writing is structural and formulaic, which makes it ideal for AI assistance. These ChatGPT prompts cover the full PM lifecycle: from writing crisp user stories that engineers can actually implement, to building prioritization frameworks for quarterly planning, to preparing data-backed presentations for leadership. Each prompt uses the frameworks PMs rely on most: RICE (Reach × Impact × Confidence ÷ Effort), MoSCoW (Must/Should/Could/Won't), Jobs-to-be-Done (the Clayton Christensen framing for unmet customer needs), OKR (the Doerr/Grove objective-and-key-result structure), and structured PRD formats. The worked example below shows how to pair the PRD prompt with the User Story prompt to surface edge cases and open questions before engineering review.

Key Takeaways

  • PM prompts should invoke NAMED frameworks: RICE for prioritization, MoSCoW for scope, JTBD for user needs, OKR for objectives. Named frameworks anchor ChatGPT to the actual PM lexicon your team uses — 'help me prioritize' produces amateur output, 'apply RICE scoring' produces reviewable analysis.
  • Run the PRD prompt and then the User Story prompt on its output. The acceptance criteria step forces an edge case and an error state for every story, which exposes gaps in the PRD before engineering reads it.
  • User story prompts must specify the 'so that' clause explicitly. 'As a [user], I want [feature]' is incomplete without the 'so that [outcome]' — that clause is what forces PMs to test whether the feature actually delivers value. Prompts skipping it produce weak stories.
  • Stakeholder update prompts should specify the audience seniority AND the desired action: 'CEO update, needs to approve budget increase' produces different output than 'engineering team update, needs to know scope changed'. Never just 'write a status update'.
“To design the best UX, pay attention to what users do, not what they say. Self-reported claims are unreliable, as are user speculations about future behavior.”
— Nielsen Norman Group — First Rule of Usability? Don't Listen to Users · nngroup.com
Worked example

Drafting a PRD for a bank-transfer retry feature with the PRD and User Story prompts

Scenario: Say you're the PM at a fintech app and failed bank transfers are your top support topic. You need a PRD for an automatic retry feature that will go through engineering, design and legal review. Paste into the Full Product Requirements Document prompt what you actually have: the support-ticket tags for failed transfers over the last quarter, five or six verbatim quotes from customer calls, the current transfer-failure rate from your dashboard, and the business objective (fewer failed-transfer tickets).

Walkthrough: Run the PRD prompt first and read the Non-goals and Open questions sections before anything else — those are where a fintech PRD usually needs legal and compliance input, so flag them for that reviewer. Then paste the user stories from the PRD into the User Story Writing prompt. Its acceptance-criteria step asks for a happy path, an edge case and an error state per story, which is where gaps show up: a retry after the account is closed, a retry that would push the account overdrawn, a user cancelling mid-retry. Add the ones you agree with back into the PRD's requirements, and ask a follow-up for any empty or error states still missing. Check every number in the draft against your own dashboard — the model will happily invent a baseline if you didn't give it one — and send the tech lead the open questions before the review meeting rather than in it.

1

Full Product Requirements Document (PRD)

Write a complete Product Requirements Document (PRD) for the following feature or product:

Feature/Product name: [name]
Product context: [brief description of the product it belongs to, the company type, and the business model]
Problem being solved: [describe the user problem or business problem in 2-3 sentences]
Target user: [specific user persona — job title, context, and pain point]
Business objective: [what business metric does this improve? e.g., activation rate, retention, revenue]

Structure the PRD with these sections:

**1. Overview**: Problem statement, proposed solution, and why now (1 paragraph each)
**2. Goals and success metrics**: 3-5 measurable success criteria with baseline and target values
**3. Non-goals**: What this feature explicitly will NOT do (to prevent scope creep)
**4. User stories**: 5-8 user stories in "As a [user], I want [action] so that [outcome]" format
**5. Functional requirements**: Numbered list of what the feature must do
**6. Non-functional requirements**: Performance, security, accessibility requirements
**7. UX/Design considerations**: Key user flows and edge cases to address
**8. Technical considerations**: Known dependencies, API requirements, or architectural notes (leave placeholders for engineering to fill)
**9. Launch plan**: Phased rollout recommendation (MVP vs V1 vs V2)
**10. Open questions**: 3-5 questions that need answers before development begins
**11. Appendix**: Relevant research, data, or competitive examples

Target length: 800-1200 words. Write for both engineering and stakeholder audiences.
Share the PRD with your tech lead before the planning meeting and ask for their top 3 technical questions. Resolving those in advance turns a contentious planning session into a smooth kickoff.
2

User Story Writing (with Acceptance Criteria)

Write a complete set of user stories with acceptance criteria for the following feature:

Feature: [describe the feature]
Product: [product name and type]
Users involved: [list the user types who interact with this feature]
Business context: [why does this feature exist? What metric does it improve?]

For each user story:
1. **User story**: "As a [specific user persona], I want [specific action] so that [specific outcome]"
   - Be specific: don't write "As a user" — name the persona ("As a free-tier user who hasn't set up their profile")
   - The action should describe intent, not implementation
   - The outcome should describe a tangible benefit, not a feature

2. **Acceptance criteria** (Given/When/Then format):
   - Given [precondition]
   - When [action taken]
   - Then [expected result]
   - Include at least 3 criteria per story: the happy path, an edge case, and an error state

3. **Story points estimate**: Suggest a relative complexity estimate (1, 2, 3, 5, 8, 13) and explain the rationale in one sentence

4. **Dependencies**: Note any other stories or systems this depends on

Write [5-8] user stories covering: the primary happy path, [2-3] alternative flows, [1-2] error/edge cases, and [1] admin or internal user story (if applicable).
Acceptance criteria that specify error states are the most commonly skipped — and the most expensive to retrofit. Engineers build what's specified; unspecified edge cases become bugs.
3

Roadmap Prioritization with RICE Framework

Help me prioritize the following product backlog items using the RICE scoring framework (Reach × Impact × Confidence ÷ Effort).

My product: [product name and brief description]
Current business priority: [e.g., improving activation rate, reducing churn, growing revenue]
Quarter timeframe: [Q1/Q2/Q3/Q4 YEAR]

Backlog items to score:
1. [Feature/initiative 1]: [one-line description]
2. [Feature/initiative 2]: [one-line description]
3. [Feature/initiative 3]: [one-line description]
4. [Feature/initiative 4]: [one-line description]
5. [Feature/initiative 5]: [one-line description]
6. [Feature/initiative 6]: [one-line description]

For each item, estimate and score:
- **Reach** (users impacted per quarter): Low/Med/High + estimated number
- **Impact** (effect on the target metric): Score 0.25 / 0.5 / 1 / 2 / 3
- **Confidence** (how certain are we): 50% / 80% / 100%
- **Effort** (person-weeks to ship): Estimate and note assumptions

Calculate the RICE score for each item and rank them.

After scoring:
1. Identify the top 3 "quick wins" (high RICE score, low effort)
2. Identify any item that ranks high in RICE but should still be delayed (strategic or technical reasons)
3. Flag items with low confidence scores that need more discovery before committing
4. Recommend a quarterly roadmap sequence based on the scores
Present RICE scores as a conversation starter, not a final verdict. Stakeholders with context you don't have (sales pipeline, legal requirements, engineering constraints) should be able to override the model — what you're preventing is intuition-only prioritization with no framework at all.
4

Competitive Analysis Brief

Write a competitive analysis brief comparing [my product] to [competitor 1] and [competitor 2] in the [product category] market.

My product context:
- Product: [name and 1-sentence description]
- Target user: [who uses it]
- Key differentiators (our hypothesis): [list 2-3]
- Current weaknesses (be honest): [list 1-2]
- Business model: [pricing and monetization]

For each competitor, structure the analysis:

**[Competitor name]**
1. Overview: What they do, who they serve, estimated scale
2. Pricing model: How they charge, entry price, and enterprise tier
3. Core strengths: What they genuinely do better than us (be fair)
4. Core weaknesses: Where they fall short or receive complaints (use review site data framing: "Users frequently report...")
5. Target segment overlap: What % of their users also look at us?
6. Recent moves: Product launches, funding, acquisitions, or strategy shifts in the last 12 months
7. Our competitive response: How should we position against this competitor in sales conversations?

**Competitive matrix**: Create a comparison table with 8-10 features/criteria and indicate which product wins each (Competitor 1 / Competitor 2 / Us / Tie)

**Strategic implications**: 3 bullet points on what this analysis means for our roadmap and positioning
Read G2, Capterra, and Trustpilot reviews for each competitor before running this prompt — paste specific complaints into the analysis for accuracy. AI-generated competitive intel is only as good as the real-world data you feed it.
5

OKR Writing for a Product Team

Help me write a set of OKRs (Objectives and Key Results) for a product team at [company type: startup / growth-stage / enterprise] working on [product type].

Company-level context:
- Company's top priority this quarter: [e.g., double ARR, launch in new market, reduce churn by 20%]
- Product team's charter: [what the product team is responsible for]
- Current key metrics baseline: [list 2-3 with current values]

Write 3 Objectives for the product team (one for each of these lenses):
1. A **customer-centric objective**: focused on user value and experience
2. A **business-results objective**: focused on revenue, growth, or efficiency metrics
3. A **team-capability objective**: focused on improving the team's speed, quality, or process

For each Objective:
- Write the Objective statement: aspirational, directional, memorable (1 sentence, no numbers)
- Write 3-4 Key Results: specific, measurable, time-bound outcomes (not activities)
  - Each KR should have a baseline value and a target value
  - Mix leading and lagging indicators (don't only use lagging metrics like revenue)
  - Include at least 1 "guardrail" KR that ensures you don't sacrifice one metric to hit another

After writing the OKRs:
- Flag any Key Results that are output-focused (what we ship) rather than outcome-focused (what changes for users/business) and suggest how to rewrite them
- Identify potential conflicts between the 3 Objectives
Use this prompt to generate a first draft, then run it by your engineering lead and a stakeholder — the friction in aligning on OKRs IS the value, not the document itself.
6

Sprint Planning Summary

Write a sprint planning summary document for the upcoming [sprint number] sprint at [company/team name]. This document will be shared with stakeholders before the sprint starts.

Sprint context:
- Sprint duration: [1 week / 2 weeks]
- Sprint goal: [1-2 sentences — what's the single most important thing this sprint achieves?]
- Team size: [X engineers, X designers, X QA]
- Total capacity: [X story points or X person-days]

Stories committed to this sprint:
1. [Story title] — [X] points — [assigned to]
2. [Story title] — [X] points — [assigned to]
3. [Story title] — [X] points — [assigned to]
[continue list]

Write the sprint planning summary with these sections:
1. **Sprint goal**: One clear sentence describing the sprint's purpose
2. **Why this sprint**: How this sprint connects to quarterly OKRs or roadmap milestones
3. **Committed scope**: A structured table of stories with owner, points, and acceptance criteria link
4. **What we're NOT doing**: Stories that were deprioritized and why (to prevent stakeholder confusion)
5. **Risks and blockers**: Known dependencies or blockers that could affect delivery
6. **Definition of done**: The criteria a story must meet to be considered complete this sprint
7. **Demo plan**: What will be shown at the sprint review and who the audience is

Keep it under 400 words. This is a communication document, not a Jira export.
Share the sprint summary 24 hours before the sprint starts — it surfaces stakeholder objections before engineering is in the middle of executing, not after. Silence before the sprint is approval; silence during is a problem.
7

User Interview Question Guide

Create a user interview guide for a product discovery session about [problem or feature area]. The interviews will be [30 / 45 / 60] minutes long and conducted with [number] users who are [describe user profile: current users / churned users / prospective users / power users].

Research goals:
1. [Primary question you're trying to answer]
2. [Secondary question]
3. [What hypothesis you're testing or trying to invalidate]

Structure the interview guide:
**Intro script (2 minutes)**: What to say to open the session, set expectations, and get recording consent
**Warm-up questions (5 minutes)**: 3 questions to understand their role, context, and relationship to [problem domain]
**Core discovery questions (20-30 minutes)**: 8-10 questions that explore their current behavior, workarounds, and pain points — NOT their opinion on features. Use Jobs-to-be-Done framing ("Tell me about the last time you...") rather than feature validation ("Would you use X if we built it?")
**Concept test section (if applicable, 10-15 minutes)**: How to introduce a prototype or concept and the 3-4 questions to ask after showing it
**Closing questions (5 minutes)**: "Is there anything important about this topic I didn't ask?" + referral ask

For each question, include:
- The question text
- A follow-up probe ("Tell me more about that" / "What did you do when that happened?")
- What you're listening for (the signal that would validate or invalidate your hypothesis)
Don't ask users what they want or whether they'd use a feature — speculation about future behavior is unreliable. Ask about a specific recent instance instead: 'When was the last time you tried to do [X]? What did you do?' Even then, memory bends, so watch them do the task when you can.
8

Stakeholder Update Email

Write a stakeholder update email summarizing the product team's progress for [week/month/quarter]. This email goes to [audience: executive team / all-hands / investors / cross-functional partners].

Update context:
- Time period covered: [dates]
- Key milestones hit: [list 2-3 specific things shipped or achieved]
- Key milestones missed: [anything that slipped and why — be direct and honest]
- Metrics movement: [list 2-3 metrics with before/after values]
- What's next: [top 2-3 priorities for the coming period]
- Key decisions needed from stakeholders: [any blockers or approvals required]

Write the email with:
1. **Subject line**: Clear and specific — not "Product Update" but "[Month] Product Update: [headline achievement]"
2. **Opening sentence**: The single most important thing that happened this period (positive or negative)
3. **Progress section**: Structured bullet points on what shipped, what moved, and what changed
4. **Metrics section**: A mini table or clear before/after comparisons — no vanity metrics
5. **Upcoming section**: What's being worked on, with realistic timelines
6. **Ask section**: Any decisions, resources, or input needed from this audience
7. **Closing**: One sentence on overall momentum or trajectory

Length: 300-400 words maximum. Stakeholders don't read long emails. If something needs more context, offer to discuss it, don't explain it inline.
The most respected PMs include bad news in stakeholder updates proactively, before being asked. A transparent update that includes a slip — with context and a recovery plan — builds more trust than a polished update that omits problems.
9

Feature Launch Announcement (Internal)

Write an internal feature launch announcement for [feature name] at [company name]. This announcement goes to [the entire company / engineering and product teams / customer-facing teams (sales, success, support)].

Feature details:
- What it does: [describe in plain language — no jargon]
- Who it's for: [user segment or customer type]
- Launch date: [date]
- Access: [all users / paid users / beta group / gated rollout]
- Key metrics we're tracking: [2-3 success metrics]
- How to access it: [URL, feature flag, navigation path]

Write the announcement with:
1. **Subject/Headline**: Punchy, specific, celebration-worthy
2. **TL;DR**: 2-3 sentences max — what launched, for whom, and what to do next
3. **What it does**: A plain-language description of the feature and the problem it solves
4. **Why it matters**: The business or user context — what does this enable?
5. **Who gets it**: Rollout details and timeline
6. **Demo/links**: Where to see it in action (Loom link placeholder, Figma link, or changelog URL)
7. **How to help**: What you need from this team — feedback, escalations, customer conversations
8. **Thanks**: Brief acknowledgment of the team that built it

Tone: energetic but precise. Avoid hype language ("game-changer," "revolutionary"). Let the feature speak for itself.
Send the customer-facing team version before the user-facing version goes live. Sales and support need to be equipped before customers ask questions.
10

Jobs-to-be-Done (JTBD) Analysis

Perform a Jobs-to-be-Done (JTBD) analysis for [product or feature] targeting [user segment]. I'll use this analysis to improve positioning, prioritize features, and guide discovery interviews.

Product: [describe the product]
User segment: [describe who this user is — job title, context, goals]
Core use case: [what do users primarily use this for today?]

For this user segment, identify:

**Functional jobs** (3-5): What tasks are they literally trying to accomplish?
Format: "When [situation], I want to [motivation], so I can [expected outcome]"

**Emotional jobs** (3): How do they want to feel while doing or after completing the task?
Format: "When [situation], I want to feel [emotion]"

**Social jobs** (2-3): How do they want to be perceived by others as a result of using this?
Format: "When [situation], I want others to see me as [perception]"

For the top 2 functional jobs:
1. **Current solutions**: How are users solving this job today? (including manual workarounds, competitor tools, or doing nothing)
2. **Struggle moments**: When does the current solution fail them?
3. **Switching triggers**: What event or frustration would make them actively search for a better solution?
4. **Success criteria**: What does "job done perfectly" look like to them?

Also produce: A one-paragraph positioning statement using JTBD language for the marketing team.
JTBD is most useful when you validate it with actual user interviews. Use this prompt to generate your hypothesis before talking to users, then verify which jobs are real pain points and which are theoretical.
11

Product Metrics Dashboard Design

Design a product metrics dashboard structure for a [product type] at a [startup / growth-stage / enterprise] company. The dashboard will be used by [product managers / executives / the whole company].

Business model: [SaaS / marketplace / consumer app / ecommerce / developer tool]
Current stage: [pre-PMF / post-PMF scaling / mature product]
Primary business goal this quarter: [specific goal]

Design the dashboard with these layers:

**North Star metric**: The single metric that best captures whether users are getting value from the product. Explain why this metric and not others.

**L1 — Health metrics** (always visible): 5-6 metrics that tell you if the product is working at a basic level. Include retention, activation, and usage frequency metrics relevant to the business model.

**L2 — Growth metrics**: 4-5 metrics that track acquisition and expansion. Include at least one leading indicator.

**L3 — Quality metrics**: 3-4 metrics for performance, reliability, and customer satisfaction (NPS, CSAT, bug rate, load time)

**L4 — Feature-level metrics**: How to track adoption of specific features — suggest a framework rather than specific metrics (which depend on features built)

For each metric:
- Name and definition
- How to calculate it
- Recommended target range or benchmark
- Warning threshold (when to escalate)
- Where to find or query this data (placeholder for your data stack)
A dashboard with 30 metrics is a dashboard with zero metrics. Ruthlessly limit L1 to 6 metrics maximum. If everything is important, nothing is — the discipline of choosing forces clarity about what you actually believe drives the business.
12

Post-Launch Retrospective

Facilitate a post-launch retrospective for [feature or product name] which launched [X weeks/months] ago. This retrospective will be run with the product team including [PM, engineering lead, designer, data analyst].

Launch context:
- Feature: [brief description]
- Launch date: [date]
- Planned success metrics: [what you said you'd achieve]
- Actual results: [what actually happened — be specific with numbers]
- Unexpected outcomes (positive or negative): [anything that happened that wasn't predicted]

Structure the retrospective agenda:

**1. Data review (15 min)**: Present the metric outcomes — what moved, what didn't, by how much vs. target. Include a "what the data tells us" interpretation.

**2. What went well (10 min)**: Specific things that worked — in execution, communication, process, or outcomes. Tie each to a reusable lesson.

**3. What didn't go well (15 min)**: Specific things that failed or underperformed. Be direct. For each, identify whether it was a: Discovery gap / Execution gap / Communication gap / External factor

**4. Biggest learning (10 min)**: The single most important thing this launch taught the team about the product, the user, or the process.

**5. Action items (10 min)**: 3-5 specific, assigned, time-bound improvements for the next launch.

Write the full retrospective document in this structure, ready to share with the team before the session. Include facilitation notes (what to ask, how to draw out quiet team members, how to prevent blame).
The most valuable retrospectives focus on process improvement, not blame assignment. If 'person X made a mistake' comes up, redirect to 'what process would have caught this earlier?' — that's actionable, blame is not.

The Numbers Behind These Prompts

Every figure below cites a primary source. Click through to verify.

3 / 2 / 1 / 0.5 / 0.25

Intercom's original RICE impact scale, from massive to minimal, alongside confidence at 100% (high), 80% (medium) and 50% (low) — anything below that it calls a "total moonshot". The Roadmap Prioritization prompt uses exactly these scales, so scores stay comparable across items.

Source: Intercom — RICE Prioritization Framework for Product Managers2026

8 hours max

The Scrum Guide timeboxes Sprint Planning to a maximum of eight hours for a one-month Sprint, and says it's usually shorter for shorter Sprints. Sprints themselves are fixed-length, one month or less. A written sprint summary from the Sprint Planning prompt is a way to send the outcome to stakeholders who weren't in the room.

Source: Scrum Guide2020

3 steps removed

NN/g's warning that when users describe their past behavior, the self-reported data is typically 3 steps removed from the truth. It's why the User Interview prompt asks about specific recent instances rather than opinions — and why watching someone do the task beats asking them about it.

Source: Nielsen Norman Group — First Rule of Usability? Don't Listen to Users2026

Common Mistakes (and How to Fix Them)

The most frequent ways these prompts go wrong — and the small phrasing changes that turn them around.

Mistake

Writing PRDs without pasting in real user research notes or telemetry

Fix

Generic PRDs read as 'someone Googled best practices'. Always paste your actual interview notes (5-8 quotes verbatim) and the telemetry that motivated the work (current funnel drop-off %, NPS for the feature area, etc). The PRD output then anchors to your evidence, and stakeholders trust it because they can see the receipts.

Mistake

Skipping the edge-case pass because the happy-path PRD looks 'done'

Fix

Engineering builds for the path you specified; QA tests for the edges. If the PRD doesn't list edge cases, engineering will discover them mid-sprint and either descope or ship bugs. Before engineering review, paste the PRD back and ask: 'List every edge case, error state, and empty state this PRD doesn't address, grouped by user story.' Then decide, case by case, which ones belong in v1.

Mistake

Using RICE without honest Confidence scores

Fix

It's tempting to give every item high confidence. RICE only works when Confidence honestly reflects uncertainty, which is why Intercom's version uses just three levels: 100%, 80% and 50%. Tell the AI: 'Use only those three Confidence levels, and for each item name the evidence behind the score'. An item with no evidence to name belongs at 50% or needs more discovery.

Mistake

Generating OKRs that are aspirational rather than measurable

Fix

'Improve user satisfaction' is not a KR. After running the OKR Writing prompt, check that every key result can be read straight from the analytics you already have, with a baseline and a target (for example, day-7 retention from [current %] to [target %]). If a KR requires building new instrumentation to measure, build the instrumentation FIRST or pick a different KR.

Mistake

Treating the RICE prioritization prompt as a 'sell the roadmap' tool rather than 'pressure-test the roadmap'

Fix

After the prompt ranks the backlog, reply: 'Now write the strongest case AGAINST each item being on the roadmap. What would a skeptical Series B board member say?' The counter-cases reveal which items have soft justifications that won't survive scrutiny — and let you either strengthen the case or drop the item before the board calls it out.

How to Use These Prompts

Copy any prompt into ChatGPT, Claude, or your preferred AI tool. Fill in every bracketed variable with your actual product context — the more specific your inputs, the more useful the output. For PRDs and OKRs, use the AI output as a first draft to review and edit with your team, not a final document to publish without review. The model doesn't know your architecture, commitments, or politics; your reviewers do. Use Prompt Anything Pro to run these prompts directly in Notion, Confluence, Linear, or any browser-based tool.

Need More Prompts?

Get personalized AI suggestions for additional prompts tailored to your specific needs.

AI responses are generated independently and may vary

Frequently Asked Questions

Run These Prompts Inside Notion, Linear & Confluence

Prompt Anything Pro lets you trigger any AI prompt directly in your browser — write PRDs and user stories without leaving your PM tools.