The Nonprofit AI Prompt Formula: Context, Audience, Goal, Voice, and Examples

Why you need a formula

Most nonprofit teams approach AI prompting informally — they write something reasonable, get something generic back, feel vaguely disappointed, and either abandon the tool or accept mediocre output. The fix is a repeatable structure. A formula doesn't constrain creativity; it provides the scaffolding that makes creativity possible. When you know what information to include before you start typing, you stop staring at a blank prompt field and start writing productive instructions. The formula is: Context + Audience + Goal + Voice + Examples.

Context: who you are and what this is for

Context tells the model what it's dealing with. "We are a Christian nonprofit serving refugee families in the Dallas–Fort Worth area" is context. "This is for our year-end fundraising appeal" is context. The more specific the context, the narrower the output range — which means less generic copy and more useful drafts. Include your organization name, who you serve, what makes your work distinct, and the specific format you're creating (email, social post, landing page, grant narrative, etc.).

Audience + Goal + Voice

Audience: who specifically is reading this, what do they already believe, and what is their relationship to your organization?

Goal: what is the one thing you want the reader to understand or do after reading this?

Voice: what tone are you aiming for — pastoral, direct, conversational, urgent, celebratory? Give the model an example sentence in your voice, or name a communicator whose style you want to approximate.

These three inputs shape everything about how the copy is written — its register, its assumptions, its energy.

Examples: the ingredient most people skip

The single most powerful addition to any nonprofit AI prompt is a real example. A specific person. A specific outcome. A real quote. A concrete detail. "Write a donor email using this story: [paste actual story here]" will produce dramatically more useful output than "write a donor email with a compelling story." The model cannot invent your specificity. It can only shape the specificity you provide. Give it the raw material, and it will work with it. Withhold the raw material, and it will invent substitutes — which is to say, it will be generic.