Audience suggestions in Advantage+ aren't targeting constraints — they're guidance signals. Understanding how to use them effectively can accelerate learning and improve performance without limiting algorithmic reach.
What Are Audience Suggestions?
In Advantage+ campaigns, audience suggestions are optional inputs that guide the algorithm's starting point:
- Interest suggestions: Categories and behaviors
- Demographic suggestions: Age and gender guidance
- Custom audience suggestions: Your own audience lists
- Lookalike suggestions: Expansion from seed audiences
Unlike traditional targeting, these are starting points, not boundaries. The algorithm can and will expand beyond them.
How Suggestions Work in Practice
Algorithm Behavior
When you provide suggestions:
- Algorithm starts by showing ads to suggested audiences
- It gathers conversion data from these initial impressions
- Patterns are identified from converting users
- Algorithm expands to similar users outside suggestions
- Over time, suggestions become less important as learning grows
This connects to how Andromeda optimizes delivery.
When Suggestions Help Most
- Early campaign learning (first 1-2 weeks)
- Lower conversion volume accounts
- Products with specific buyer profiles
- New accounts without historical data
When Suggestions Matter Less
- High conversion volume (200+ weekly)
- Mature campaigns with strong learning
- Broad-appeal products
- Accounts with rich historical data
Types of Effective Suggestions
Custom Audience Suggestions
Your strongest signal option. Suggest based on:
- High-value purchasers: Customers with highest LTV
- Repeat purchasers: Engaged, loyal customers
- Recent converters: Fresh conversion patterns
- Engaged visitors: Multi-session or long-duration visitors
Lookalike Suggestions
Lookalikes as suggestions provide expansion patterns. See our audience strategy guide:
- 1% lookalike from best customers
- Value-based lookalikes when available
- Lookalike from specific conversion events
Interest Suggestions
If using interest suggestions:
- Choose interests proven in manual campaigns
- Focus on 3-5 highly relevant interests
- Avoid broad interests that apply to everyone
- Consider competitor brands or related products
Demographic Suggestions
- Only use if buyer profile is clearly defined
- Don't restrict unless data supports it
- Consider suggesting ranges rather than narrow bands
- Gender suggestions only for gender-specific products
Strategic Approaches
Approach 1: No Suggestions (Pure Broad)
Best when:
- High conversion volume (200+ weekly)
- Broad-appeal products
- Strong creative that self-selects audience
- You want maximum algorithmic freedom
Approach 2: Light Suggestions
Add 1-2 high-quality suggestions:
- Best customer custom audience as pattern seed
- One or two proven interests
- Best for moderate volume accounts (50-200 weekly conversions)
Approach 3: Rich Suggestions
Multiple suggestion types combined:
- Custom audiences + lookalikes + interests
- Best for new accounts or niche products
- More guidance for algorithm learning phase
- May limit eventual scale
Testing Suggestion Strategies
A/B Test Framework
Test different suggestion approaches. See our testing methodology guide:
- Campaign A: No suggestions
- Campaign B: Custom audience suggestion only
- Campaign C: Interest + lookalike suggestions
What to Measure
- Learning phase exit speed
- New customer CPA after learning
- ROAS at week 2 vs week 4
- Reach and frequency patterns
Interpretation
If suggestions underperform pure broad:
- Your suggestions may be too narrow
- Algorithm has enough signal without guidance
- Consider removing suggestions
If suggestions outperform pure broad:
- Algorithm benefits from directional guidance
- Test different suggestion combinations
- Consider adding more quality suggestions
Common Mistakes
Over-Suggesting
Adding too many suggestions can narrow the algorithm's initial exploration. Start light and add only if needed.
Low-Quality Suggestions
Suggesting small or low-quality audiences gives poor signals. Use audiences with 1000+ users and proven conversion patterns.
Treating Suggestions as Targeting
Suggestions guide, not constrain. If you need hard targeting boundaries, use traditional campaigns instead.
Never Testing Alternatives
Don't assume suggestions always help. Test with and without to find your optimal approach.
How ROASPIG Helps
Audience suggestions work best when paired with creative that resonates with the algorithm's selected audiences. ROASPIG supports this:
- Audience-Aligned Creative: Generate creative that appeals to your suggested audience profiles
- Self-Selecting Design: Creative that attracts right users regardless of suggestion strategy
- Testing Support: Generate identical creative across suggestion test variants
- Performance Analysis: Understand which creative concepts perform with different suggestion strategies
- Persona Variations: Create variations targeting different buyer profiles for suggestion alignment
The Bottom Line
Audience suggestions in Advantage+ are optional guidance, not required targeting. They help most during learning phases and for lower-volume accounts. High-volume accounts with broad-appeal products often perform equally well without suggestions.
Test different approaches for your specific situation. Start light, measure impact, and adjust based on data rather than assumptions.
Frequently Asked Questions About Advantage+ Audience Suggestions
No, suggestions are optional. They can help during learning phases and for lower-volume accounts, but high-volume accounts often perform equally well without them. Test both approaches to find what works for your situation.
Custom audiences based on your best customers are typically most effective because they provide strong conversion patterns. High-value purchasers or repeat customers give the algorithm quality signals to expand from.
No, suggestions are starting points, not boundaries. The algorithm will expand beyond your suggestions to find additional converting users. They guide initial learning but don't permanently constrain reach.
Start light — one or two high-quality suggestions (best customer audience, proven interest). Over-suggesting can narrow initial exploration. Add more only if testing shows benefit. Many accounts perform best with minimal or no suggestions.
Only if you have proven interests from manual campaigns. Choose 3-5 highly relevant interests rather than broad categories. Avoid interests that apply to everyone. Custom audience suggestions typically provide stronger signals than interest suggestions.