Use audience generation when you know the campaign goal but do not want to build every filter by hand. Marve turns your prompt into one or more audience suggestions for the audience builder.

How audience generation works

Prompt
  |
  v
Marve maps intent to available fields
  |
  v
Marve validates values, joins, and counts
  |
  v
Audience suggestion appears in Studio
  |
  v
You open, review, edit, save, and export

Generated suggestions do not automatically create or export audiences. You decide what to save and activate.

Write a strong audience prompt

Include these details when you can:

• Objective: what you want the campaign to accomplish.
• Population: product tier, geography, account type, lifecycle stage, or customer type.
• Signals: behaviors, transactions, engagement, support activity, or model scores.
• Timeframe: the period Marve should evaluate.
• Exclusions: customers or accounts that should not be included.

Generate an audience

1. Open Studio.
2. Select the Dataset Group for the audience.
3. Add any relevant goals in the prompt controls.
4. Enter a specific audience prompt.
5. Send the prompt.
6. Review the generated audience suggestions.
7. Open a suggestion in the audience builder.
8. Review the filters, joins, and audience size.
9. Save the audience when it matches your intent.

⚠️

Screenshot check

Replace this screenshot if the current generated-audience review experience has changed.

Review before you activate

Before saving or exporting a generated audience:

• Confirm the audience size is plausible.
• Review each filter and exclusion.
• Check that dates match your intended timeframe.
• Confirm Marve used the right dataset relationships.
• Spot-check the logic with a teammate when the campaign is high impact.

Iterate on the suggestion

Use specific follow-ups to refine the result:

• "Add customers with at least two purchases in the last 90 days."
• "Exclude anyone who received a winback email in the last 30 days."
• "Limit this to customers in California and New York."
• "Create two more variants: one for high spend and one for high engagement."
• "Explain which fields you used for churn risk."

Common mistakes

Congratulations! You can now generate and refine audiences with AI Studio. Continue to Audience Discovery for suggested audiences on Home.