Writing a Product Requirement Document (PRD) is a crucial step in defining what a product should do, how it should work, and the constraints it must operate within.

Traditionally, writing a PRD requires a structured approach, deep product knowledge, and collaboration with stakeholders. But with AI tools like ChatGPT, you can accelerate this process by generating drafts, refining details, and structuring requirements more efficiently.

In this post, we’ll cover:

• How to use ChatGPT to generate a PRD

• How to provide context for more accurate responses

• How to refine AI-generated output into functional and non-functional requirements

• How to convert requirements into user stories with acceptance criteria

Getting Started: Providing the Right Context

ChatGPT can generate a PRD, but the quality of the output depends on the input. Before asking ChatGPT to draft a PRD, provide the following details:

Product Overview

Start with a high-level description:

• What is the product or feature?

• Who is the target audience?

• What problem does it solve?

Example prompt:

“I need a PRD for a new feature that allows SMB merchants to generate invoices through our payment portal. The target users are small business owners who need an easy way to create, send, and track invoices. The goal is to improve cash flow by making it easier for merchants to get paid.”

Business Goals & Success Metrics

AI needs to understand what success looks like. Include:

• Key business objectives (e.g., increase adoption, reduce churn)

• Metrics for success (e.g., % of merchants using the feature, average time to send an invoice)

Example prompt addition:

“The goal is to have at least 50% of our active merchants using the invoicing feature within six months. A successful outcome would be reducing payment collection time from an average of 15 days to 10 days.”

Key Features & Scope

List what you expect in scope and any exclusions.

Example prompt addition:

“The invoicing feature should include the ability to create, customize, and send invoices, accept online payments, and track status (paid, unpaid, overdue). However, advanced accounting integrations will be out of scope for this release.”

Generating the PRD with ChatGPT

Once you’ve provided the right context, you can now ask ChatGPT to generate a first draft.

Prompt Example:

“Using the details I provided, create a Product Requirement Document (PRD) for our invoicing feature. Include sections for Overview, Business Goals, Functional Requirements, Non-Functional Requirements, and Open Questions.”

Example Output (Excerpt)

Product Requirement Document (PRD) – Invoicing Feature

1. Overview

The invoicing feature will allow SMB merchants using our payment portal to generate, send, and track invoices. This feature aims to improve cash flow by simplifying the process of getting paid.

2. Business Goals & Success Metrics

• Increase merchant engagement with our platform by adding invoicing capabilities.

• Ensure at least 50% adoption among active merchants within 6 months.

• Reduce the average time to receive payments from 15 to 10 days.

3. Functional Requirements

• Merchants can create an invoice by entering customer details, line items, and payment terms.

• Ability to customize invoice branding (logo, colors).

• Send invoices via email with a payment link.

• Track invoice status (draft, sent, paid, overdue).

4. Non-Functional Requirements

• System must be able to handle up to 10,000 invoices per hour.

• The UI should adhere to existing design system guidelines.

5. Open Questions

• Should we allow invoice reminders to be automated?

• Will merchants need the ability to apply discounts or taxes?

Refining Requirements: Functional vs. Non-Functional

Once ChatGPT has provided a draft, you’ll need to refine it by separating functional and non-functional requirements.

Functional Requirements

Non-Functional Requirements

If ChatGPT’s response isn’t detailed enough, ask follow-up questions like:

“Can you provide more detailed functional requirements for invoice customization?”

Converting Requirements into User Stories

Once the PRD is structured, the next step is to create user stories with acceptance criteria.

Example Prompt:

“Convert the functional requirements into user stories using the ‘As a [user], I want to [action] so that [benefit]’ format. Include acceptance criteria.”

Example Output:

User Story 1: Create an Invoice

As a merchant, I want to create an invoice by adding customer details and line items so that I can bill my customers efficiently.

Acceptance Criteria:

Merchant can input customer name, email, and business details.

Merchant can add multiple line items with quantity and price.

The system calculates total cost, including taxes if applicable.

User Story 2: Send an Invoice via Email

As a merchant, I want to send an invoice via email so that my customer receives it instantly and can make a payment.

Acceptance Criteria:

The invoice is sent to the provided email address.

The email contains a payment link.

Merchant receives a confirmation that the invoice was sent.

Iterating & Finalizing the PRD

Once you have the PRD, functional requirements, and user stories, the final steps include:

• Reviewing with stakeholders – Validate with engineering, design, and business teams.

• Breaking down technical details – Work with engineering to ensure feasibility.

• Refining acceptance criteria – Make them measurable and testable.

Final Thoughts

ChatGPT is a powerful tool for drafting and refining Product Requirement Documents (PRDs). By providing clear context, iterating on AI-generated outputs, and validating details with stakeholders, you can streamline the process while maintaining high-quality requirements.