Updated: 31 October 2026

One-page product requirements doc

This prompt writes a one-page PRD for a feature from the problem and evidence, users, goals and success metrics and constraints, including non-goals and the main user flow in steps.

text-prd-one-pager — chatgpt — 45 words

Write a one-page product requirements document for [FEATURE].

Problem and evidence: [PROBLEM]
Users: [USERS]
Goals and success metrics: [METRICS]
Constraints: [CONSTRAINTS]

Include non-goals, the main user flow in steps, key requirements as a prioritised list, risks, open questions and a launch plan in three lines.

Edit in the builder Copy & open ChatGPT ↗Copy & open Gemini ↗Copy & open Claude ↗

Share it early with engineering; they will spot the hard parts before you commit.

Get a better result

What you will get

A one-page PRD with the problem and evidence, users, goals and success metrics, non-goals, the main user flow in steps, prioritised requirements, risks, open questions and a three-line launch plan.

Why it works

A PRD keeps product, design and engineering aligned on why something is being built. The prompt asks for evidence, non-goals and metrics because those are where teams disagree later. Keeping it to one page forces priorities, and sharing it early, as the tip says, lets engineering raise the hard parts before anyone commits.

Common mistake

The evidence section is an opinion.

Fix: Paste real signals, such as support ticket counts or quotes from user calls. A PRD without evidence is easy to argue with.

Also watch for

Every requirement is marked high priority. Ask for must, should and could labels, with no more than a third marked must.
The success metric cannot be measured yet. Check you already track it. If not, add tracking as a requirement.

Try a variation: experiment brief

Turn it into a brief for a two-week experiment: the hypothesis, the smallest version to build, the metric and the result that would make you continue.

Still not right? Reply with this

Rewrite the requirements as must, should and could, keep the musts to the smallest set that solves the problem, and move the rest to a later phase.