How to write the best discovery autopilot prompt

Last updated: February 23, 2026

What is the Discovery Autopilot feature?

It is a new search mechanism to ensure high-quality results in your creator discovery process. Based on the user-provided search inputs (e.g., lookalike or filters, prompt, and content style), the autopilot will first search the Meta Creator Marketplace and, after that, use AI to analyse profiles and return the best results.

Best Practices for Effective Prompts:

  1. Combine multiple criteria: "Fitness creators targeting women 25-35, posting authentic home workout content with high engagement."

  2. Use descriptive language: Instead of "fashion," try "sustainable fashion with thrift shopping and styling tips."

  3. Specify content characteristics: "Educational skincare content from dermatologists or estheticians, featuring product recommendations."

  4. Include audience context: "Parent lifestyle content for millennial moms, showing real-life parenting moments."

  5. Consider authenticity: "Authentic, unfiltered travel content showing local experiences and hidden gems."

  6. Think about brand safety: "Family-friendly cooking content without controversial topics."

Example of an Excellent Prompt:

"Find micro-influencers (10k-100k followers) creating authentic home cooking content for busy parents. Looking for creators who feature quick meal prep, kid-friendly recipes, and kitchen organization tips. Content should have a casual, relatable style with high authenticity scores. Target audience: parents ages 25-40, primarily female. High engagement rates preferred."

This prompt leverages:

  • Metrics (follower range, engagement)

  • Content topics (cooking, meal prep, recipes)

  • Objects/environments (kitchen)

  • Audience (parents, age, gender)

  • Content style (casual, authentic)

  • Purpose (helping busy parents)

✅ What you CAN and SHOULD search for:

1. Content Topics & Themes

The video analysis extracts topics, an overview, and a purpose from each post.

Examples:

  • "Creators who post about sustainable fashion."

  • "fitness content focused on home workouts."

  • "tech reviews and unboxing videos."

2. Visual Elements & Objects

The system identifies objects and environments in videos.

Examples:

  • "Creators who feature outdoor environments."

  • "Content showing cooking equipment and kitchen setups."

  • "Videos featuring pets or animals"

3. Brands & Product Categories

The analysis detects brands mentioned or shown in content.

Examples:

  • "Creators who work with athletic wear brands."

  • "Content featuring beauty and skincare products."

  • "Posts mentioning sustainable brands."

4. People & Professions

The system identifies people types (professions, demographics) in videos.

Examples:

  • "Doctors or healthcare professionals sharing medical tips."

  • "Athletes demonstrating training techniques."

  • "Working parents sharing lifestyle content."

5. Content Style & Authenticity

Examples:

  • "Authentic, unedited content with high authenticity scores."

  • "Educational content with a professional tone."

  • "Humorous, casual content."

6. Audience & Demographics

Available from profile demographics.

Examples:

  • "Creators targeting young adults 18-24."

  • "Content for female audiences."

  • "Creators with Spanish-speaking followers."

7. Engagement & Performance Metrics

Examples:

  • "Creators with high engagement rates."

  • "Consistent posters who upload 3+ times per week."

  • "Micro-influencers with 10k-100k followers."

8. Bio & Profile Information

  • "Creators who mention 'plant-based' in their bio."

  • "Profiles located in Los Angeles."

  • "Creators in the lifestyle and wellness categories."

9. Advertising & Commercial Content

Examples:

  • "Creators who post non-ad content primarily."

  • "Profiles with authentic product integrations."

10. Risk Awareness

The system identifies potential content risks.

Examples:

  • "Content without controversial topics."

  • "Brand-safe creators."

❌ What Users SHOULD NOT search for:

1. Specific Creator Names

The autopilot is for discovery, not lookup. Use regular search for known creators.

Example:

  • "Find @username"

2. Exact Numeric Values

Use ranges and relative terms instead: "50k-100k followers" or "mid-tier influencers"
Example:

  • "Exactly 50,000 followers" -

3. Platform Technical Features

Focus on content characteristics, not platform mechanics.

Example:

  • "Creators who use Instagram Reels filters" -

4. Temporal/Time-Based searches Without Context

The system analyzes content patterns, not real-time posting schedules.

Example:

  • "Creators who posted yesterday"

5. Off-Platform Behavior

Only the analyzed Instagram data is available.

Example:

  • "Creators with YouTube channels" or "TikTok presence."

6. Personal Contact Information

Focus on content fit, not contact logistics.

Example:

  • "Creators with public email addresses"

7. Vague or Overly Broad Requests

Be specific about content types, topics, style, or audience.

Examples:

  • "Good creators" or "Popular people"