Skip to contentComing soon: SprintProof for ChatGPT and Codex

Free template

Product requirements document (PRD) template

Ten sections that get a feature from idea to something engineering can estimate: the problem, who it is for, what done means, what is out, and the questions to settle before anyone writes code. Each section says what to write and what reviewers push back on.

Markdown pastes into Notion, Google Docs and Confluence. No sign-up.

PRD generator

Fill it in here, copy it out.

Type into the form and the template below fills in as you go. Copy the Markdown or download it as a .md file. It runs in your browser and nothing you type is sent anywhere.

PRD generator
runs in your browser · nothing is sent

The situation today, without the solution.

One per line. “Admin: invites people” fills the table’s second column.

One per line, each something a tester can check.

One requirement per line.

One per line, with a reason.

One per line. Owners and dates stay as placeholders.

Preview (Markdown)

# [Feature name]: product requirements

**Owner:** [name] · **Status:** Draft · **Last updated:** [date]
**Reviewers:** [product] · [engineering] · [design] · [area owners]

## 1. Summary

[Summary: what you are building, for whom, and why now, in three sentences or fewer. Write it last.]

## 2. Problem

[The situation today, without the solution: who has the problem, what they do instead, the evidence.]

## 3. Users and roles

| Who | What they can do | What they must not be able to do |
| --- | --- | --- |
| [role] | [actions] | [limits] |

## 4. Done when

- [ ] [A user can ...]

**Measured after launch:** [metric], baseline today: [value or "no baseline yet"]

## 5. Not now

- [Out of scope item]: [reason]

## 6. Requirements

| # | Requirement | Layer | Priority | Best (days) | Likely (days) | Worst (days) |
| --- | --- | --- | --- | --- | --- | --- |
| R1 | [requirement] | Frontend / Backend / Data / Cross-cutting | Must / Should / Could | | | |

## 7. Edge cases and open questions

| Question | Owner | Decide by | Answer |
| --- | --- | --- | --- |
| [question] | [name] | [date] | |

## 8. Risks and dependencies

- **Risk:** [what could go wrong] · **Mitigation:** [what you will do] · **Owner:** [name]
- **Depends on:** [team or system] for [what], by [date]

## 9. Estimate

- **Total:** [best] to [worst] days, likely [likely]
- **Widest spread:** [requirement]: [why it is uncertain]
- **Date at 80% confidence:** [date]

## 10. Sign-off

| Role | Name | Decision | Conditions |
| --- | --- | --- | --- |
| Product | [name] | Approved / Changes requested / Pending | |
| Engineering | [name] | | |
| Design | [name] | | |
| [Area owner] | [name] | | |

---
Made with the free PRD generator from SprintProof: https://sprintproof.com/templates/prd-template

The template

Ten sections, each with what to write.

A PRD is read by product, engineering, design and whoever owns the areas it touches. Write each section for the person who will push back on it.

prd-template.md
blank, with guidance
  1. 01

    Summary

    What you are building, for whom, and why now, in three sentences or fewer.

    • What changes for the user?
    • Who asked for it, or what data points to it?
    • Why this quarter rather than next?

    TipWrite it last. If it will not fit in three sentences, the scope is not settled yet.

  2. 02

    Problem

    The situation today, without the solution. Reviewers judge the solution against this.

    • Who has the problem, and how often?
    • What do they do instead today?
    • What evidence do you have: tickets, interviews, usage?

    TipLink the evidence rather than summarising it. A problem with no source reads as an opinion.

  3. 03

    Users and roles

    Every kind of person the feature touches, including the ones who only receive something.

    • Who starts the flow, and who is on the other end?
    • Which roles may do what?
    • Who must not be able to do it?

    TipName the permission rules here. They are the part most often left for engineering to guess.

  4. 04

    Done when

    Outcomes a tester can check. This is what the plan is reviewed and signed off against.

    • What can a user do when this ships?
    • Under which conditions must it still work?
    • What will you measure afterwards, and what is the baseline today?

    Tip“Improve onboarding” cannot be tested. “An admin can invite five people in one go” can.

  5. 05

    Not now

    What is deliberately out of scope for this release, each with a reason.

    • What will people assume is included?
    • What did you cut to make the date?
    • What comes in a later release?

    TipThis is the cheapest scope-creep control you have. Anything not listed will be asked for in review.

  6. 06

    Requirements

    What has to exist, grouped by layer: interface, API, data and the rules that cut across them.

    • Which screens, endpoints and tables change?
    • Which states does each screen need: empty, loading, error, no access?
    • Which items are must, should or could?

    TipDescribe behaviour, not implementation. Leave the how to the technical design.

  7. 07

    Edge cases and open questions

    What you do not know yet, each with an owner and a date to decide by.

    • What happens when the input already exists, expires or is wrong?
    • Which decisions belong to another team?
    • What would change the estimate if answered differently?

    TipAn open question without an owner turns into scope creep in the middle of the sprint.

  8. 08

    Risks and dependencies

    Anything outside the team that can slow the work: other teams, vendors, migrations, reviews.

    • Which part is new to the team?
    • Who else has to ship or approve something?
    • Does it need a security, legal or privacy review?

    TipPut the riskiest part first in the plan and spike it, rather than finding out in week two.

  9. 09

    Estimate

    A range per requirement and in total, so everyone sees how far the bad case is from the likely one.

    • Best, likely and worst case per part, in days
    • Which parts have the widest spread, and why?
    • The date you would commit to at 80% confidence

    TipNever one number. A single date is read as a promise, and the worst case is where the conversation should start.

  10. 10

    Sign-off

    Who approves the plan before work starts, and what each of them is approving.

    • Product, engineering and design
    • Owners of the areas it touches: billing, security, support
    • Their decision and any condition attached

    TipAsk the area owners early. Billing and security rarely say no, but they often say “not like that”.

Example

A PRD for team invitations, filled in.

A real-sized feature: workspace admins invite colleagues by email. This is the same plan SprintProof’s sample workspace uses, written up as a PRD.

team-invitations-prd.md
example, filled in
INV · PRD · OWNER MARIA K. · IN REVIEWTeam invitations: product requirements

1 · Summary

Owners and admins invite colleagues by email with a role, and track who has not joined. Today they have to ask us to add each person by hand. The accept flow touches sign-in, so it is planned first.

2 · Problem

Workspace owners and admins need to bring colleagues in without asking us to add them by hand. They invite people by email, choose a role, see who has not accepted yet, and can resend or revoke. Invitees get a link that works for seven days and lands them in the right workspace.

3 · Users and roles

  • Owners and adminsCan: Invite by email, pick a role, resend and revokeCannot: Invite a role above their own
  • MembersCan: See who is in the workspaceCannot: Invite anyone
  • InviteesCan: Accept once, within 7 days, as the invited addressCannot: Reuse or forward the link to join

4 · Done when

  • An owner or admin can invite one or many people by email and pick a role
  • An invitee can accept from the email link when signed out, signed in, or signed in as someone else
  • Admins see pending invites and can resend or revoke them
  • Invites expire after 7 days and work once

5 · Not now

  • Domain auto-join: needs verified domains first
  • SSO and SCIM provisioning: comes with the SSO work in a later release
  • CSV import of invitees: pasting many addresses at once covers it for now

6 · Requirements, in dev-days

RequirementLayerOwnerBestLikelyWorst
R1Invite modalEmail input, role picker, paste many addresses at onceFrontendSam O.11.53
R2Pending invites listSee, resend and revoke invites nobody has accepted yetFrontendLena V.123
R3Accept-invite pageLogged out, logged in, wrong account, expired linkFrontendSam O.1.524
R4Invitations APICreate, list, revoke, acceptBackendKen L.235
R5Invite tokensSigned, single-use, expire after 7 daysBackendJonas T.0.512
R6Invite emailTemplate, sent through the job queueBackendKen L.11.53
R7invitations tableNew table and migration, index on workspace + emailDataKen L.0.511.5
R8Permission rulesWho may invite which role: owners vs. adminsCross-cuttingJonas T.11.53
R9TestsAPI tests and one end-to-end run: invite, accept, first loginCross-cuttingLena V.1.523
R10Rate limit invite sendsfrom review · MAIL-0220 invites per admin per hour, then a clear errorBackendKen L.0.50.51
Total10.51628.5

7 · Open questions

  • What if the invited email already has an account?Jonas T. · Tue
  • What happens when an invite goes over the plan’s seat limit: block, warn or charge?Tom A. · Wed
  • Can anyone with the link join, or only the invited address?Jonas T. · Tue
  • How does the admin find out an invite bounced?Ken L. · Thu

8 · Risks and dependencies

  • Accepting an invite touches sign-in, usually the riskiest part.Spike the accept flow first · Ken L.
  • Invite emails go to addresses a user typed, so they can be used to send spam from our domain.Rate limit sends (added from review) · Ken L.
  • The seat-limit answer changes the modal and needs billing’s decision.Decide by Wed, before the modal is built · Tom A.

9 · Estimate

10.5 to 28.5 dev-days, likely 16. Widest spread: Invitations API (2 to 5), because create, revoke and accept all touch the same tokens.

The open questions on existing accounts and seat limits each add a part if the answer is “yes”, so the range widens until they are decided.

10 · Sign-off

  • Product · Maria K.ApprovedScope matches the brief. Out-of-scope list looks right.
  • Tech lead · Jonas T.Pending
  • Design · Priya R.Changes requestedAccept page needs a “signed in as someone else” state.
  • Billing · Tom A.PendingSeat limit policy: block, warn or charge?

Comparing PRD tools? See ChatPRD alternatives.

How to use it

How to fill it in, step by step.

A PRD is a conversation starter, not a contract. Most of its value comes from the review: the questions people ask and the scope you agree to drop.

  1. 01Start with the problem and done-when. Leave the summary blank. Write the problem with links to the evidence, then the outcomes a tester could check.
  2. 02List what is not in this release. Everything people might assume is included, each with a reason. It is easier to agree on now than in review.
  3. 03Break the requirements down by layer and give each one a best, likely and worst case. Ask the engineers who will build it, not the one who knows the code best.
  4. 04Walk the edge cases. Existing accounts, expired links, missing permissions, empty and error states. Each one is either a requirement or an open question with an owner.
  5. 05Send it for review to product, engineering, design and the owners of every area it touches. Record each sign-off and its conditions in the last section.
  6. 06Write the summary last, once the scope has settled. Keep the PRD next to the tickets so people can check a question against it.

Skip the blank page

Let SprintProof draft it with you.

Describe the feature and the planner asks the questions a reviewer would, then drafts the brief, done-when, not-now and parts with ranges. The check library adds the edge cases teams usually find mid-sprint.

Turn this into a reviewed plan

FAQ

PRD questions

What is a PRD?
A product requirements document describes what a feature has to do and why: the problem, the users, the outcomes that count as done, what is out of scope and the open questions. It leaves how to build it to the technical design.
How long should a PRD be?
As short as the feature allows. For a feature of a few weeks, one to three pages is usual. If a section has nothing to say, write “none” so reviewers know it was considered.
What is the difference between a PRD and a technical design doc?
The PRD says what and why, and is owned by product. The technical design doc says how: architecture, data model, alternatives and rollout, and is owned by engineering. The design doc should answer the PRD’s requirements and open questions.
Who should write the PRD?
Usually the product manager, with the tech lead and designer reviewing early drafts. On teams without a product manager the engineering lead often writes it. What matters is that one person owns it.
Is the PRD generator private?
Yes. It runs in your browser and nothing you type is sent anywhere or saved. It uses no AI. Copy the Markdown or download the file, and close the tab when you are done.
How do I use the template in Notion or Google Docs?
Copy it with the button on this page and paste it: Notion turns the Markdown into headings and lists. In Google Docs, turn on Tools, Preferences, “Enable Markdown” first, then use Edit, “Paste from Markdown”. Or download the .md file and import it.

Turn this into a reviewed plan with SprintProof.

Describe the feature and SprintProof turns it into a plan: checked against a library of questions reviewers ask, estimated as a range with 50, 80 and 90% dates, and signed off before anyone starts.