Free PRD template

This is a free product requirements document (PRD) template in Markdown, with a guidance comment under every section. Use it when a feature is big enough that product, design and engineering need one written agreement on the problem, the evidence, the scope and how you will know it worked, before anyone starts building.

Updated

How do you use this template?

  1. Copy the template below into your doc tool, wiki or repository. It is plain Markdown, so it pastes cleanly into Notion, Confluence, Google Docs or a .md file.
  2. Fill in Problem and Evidence first, before any solution. If you cannot fill in Evidence, do more research before writing the rest.
  3. Write Non-goals early. It is easier to agree on what you are not doing before the requirements list grows.
  4. Number every requirement and give each P0 and P1 testable acceptance criteria.
  5. Review with design and engineering, record open questions with owners, and delete the guidance comments once a section is done.

The PRD template

prd-template.md
# PRD: [Feature name]

| Field | Value |
|---|---|
| Owner | [PM name] |
| Reviewers | [Design lead], [Engineering lead] |
| Status | Draft / In review / Approved / Shipped |
| Last updated | [YYYY-MM-DD] |
| Links | [Design file], [Research], [Epic or project] |

## 1. Summary
<!-- Two or three sentences: what we are building, for whom, and why now.
     Someone who reads only this section should understand the decision. -->

## 2. Problem
<!-- Describe the problem, not the solution. Who has it, when does it happen,
     and what does it cost them (time, money, risk, a workaround)?
     If this section mentions your feature, rewrite it. -->

## 3. Evidence
<!-- Link the research that shows the problem is real and worth solving.
     Prefer quotes from more than one source. Say how strong the evidence is. -->

| Evidence | Source | Strength |
|---|---|---|
| "[Verbatim customer quote]" | [Interview 4, Finance lead, 2026-10-02] | Repeated across 4 of 8 interviews |
| [Ticket volume or usage data] | [Support export, Q3] | Behavioral |

## 4. Target users
<!-- Primary user, secondary users, and who this is explicitly NOT for. -->
- Primary:
- Secondary:
- Not for:

## 5. Goals and success metrics
<!-- What will change if this works? Each goal needs a metric, a baseline
     and a target, plus when you will measure it. -->

| Goal | Metric | Baseline | Target | Measured when |
|---|---|---|---|---|
| | | | | |

## 6. Non-goals
<!-- What we are deliberately not doing in this release, and why.
     This is the section that stops scope creep. Never leave it empty. -->
-

## 7. User flows
<!-- The main flow step by step, then edge cases and error states:
     empty data, no permission, offline, invalid input, partial failure. -->
1.
2.
3.

Edge cases:
-

## 8. Requirements
<!-- One row per requirement. Priorities: P0 = must ship, P1 = should ship,
     P2 = nice to have. Every P0 and P1 needs testable acceptance criteria,
     ideally in Given / When / Then form. -->

| ID | Requirement | Priority | Acceptance criteria |
|---|---|---|---|
| REQ-001 | | P0 | Given ..., when ..., then ... |
| REQ-002 | | P1 | |

## 9. Non-functional requirements
<!-- Performance, security, privacy, accessibility, localization, limits. -->
-

## 10. Design
<!-- Link the designs. Note any decisions the designs do not make obvious. -->

## 11. Dependencies and risks
<!-- Other teams, services or vendors this relies on, and what could go wrong. -->

| Risk or dependency | Impact if it goes wrong | Mitigation |
|---|---|---|
| | | |

## 12. Assumptions and open questions
<!-- Assumptions you are making but have not validated, and questions that
     still need an owner and a date. Resolve high-impact ones before build. -->

| ID | Assumption or question | Impact if wrong | Owner | Due |
|---|---|---|---|---|
| ASM-001 | | High / Medium / Low | | |

## 13. Launch plan
<!-- Rollout (flag, beta, percentage), who needs to know (support, sales,
     docs), and how you will announce it. -->

## 14. Changelog
<!-- Date and one line for every meaningful change to this document. -->
- [YYYY-MM-DD] Created.

Common mistakes when writing a PRD

  • Describing the feature in the Problem section. That rules out better solutions before anyone considers them.
  • Leaving Evidence empty. Without it, every review reopens whether the problem is real.
  • Untestable requirements. "Fast", "simple" and "intuitive" are not acceptance criteria.
  • No edge cases. Empty states, errors and permissions are where most rework comes from.
  • Letting it go stale. Update the changelog when decisions change, or people stop trusting the document.

How Praxiom AI fills this in for you

Praxiom AI drafts a PRD from your synthesized research, using insights and their source quotes as the evidence. You edit the draft in the app, export it to PDF or Markdown, and split it into scoped tickets in Linear, Jira, GitHub. See PRD generation or sign up.

Frequently asked questions

Is this PRD template free?

Yes. Copy it, change it and use it for any product, with no sign-up.

What sections should every PRD have?

At minimum: problem, evidence, target users, goals with success metrics, non-goals, user flows and requirements with acceptance criteria. Risks, open questions and a launch plan matter more as the feature gets bigger.

Can I hand this PRD to an AI coding agent?

It is a good start, but agents need more precision than people do. Use the spec packet template to add constraints, a test plan and likely files, and to resolve open questions first.

How is a PRD different from a one-pager?

A one-pager pitches the problem and the bet to get agreement to proceed. A PRD defines what will be built in enough detail to build it. Many teams write the one-pager first, then expand it into this template.

Run your product discovery on autopilot.

Bring your interviews, tickets and surveys. Get scored insights, a PRD and tickets.