Lookalike audiences are one of the most powerful targeting tools in mobile advertising. By modeling from your best users, you can find similar high-value prospects at scale. Here's how to maximize their effectiveness.
How Lookalike Audiences Work
Platforms analyze your seed audience's characteristics—demographics, behaviors, interests—and find new users who share similar patterns but aren't yet customers.
The Modeling Process
- You provide a seed audience (customer list or pixel data)
- Platform identifies common traits among seed users
- Algorithm finds similar users not in your seed
- Audience is created for targeting
🎯 Quality Tip
The quality of your lookalike directly depends on the quality of your seed. A seed of 1,000 high-LTV users will outperform a seed of 100,000 random installers every time.
Seed Audience Selection
Your seed audience determines lookalike quality. Choose wisely.
High-Quality Seed Options
- High-LTV customers: Users with top 20% lifetime value
- Purchasers: Users who made at least one purchase
- Engaged users: D30+ retained users
- Subscription converters: Users who converted to paid
- Multi-purchasers: Repeat buyers
Seed Size Guidelines
- Minimum: 100-500 users (platform dependent)
- Optimal: 1,000-10,000 users
- Diminishing returns: Above 50,000 users
Seed Quality vs Quantity
When choosing between more users or better users, choose better:
- 1,000 whales > 10,000 installers
- 500 subscribers > 5,000 trial starters
- Recent users > historical users
Lookalike Expansion Size
Most platforms let you choose how closely matched your lookalike should be.
Expansion Trade-offs
- 1% lookalike: Highest similarity, smallest reach, best quality
- 5% lookalike: Moderate similarity, moderate reach
- 10% lookalike: Lower similarity, largest reach, lower quality
When to Use Each Size
- 1-2%: High-value campaigns, premium pricing, limited budget
- 3-5%: Balanced approach, most common starting point
- 6-10%: Scale campaigns, broad awareness, large budgets
"Start narrow, then expand. It's easier to scale from a 1% lookalike to 5% than to fix a poor-performing 10% audience."
Advanced Lookalike Strategies
Layered Lookalikes
Create multiple lookalikes and layer them:
- Primary: High-LTV purchasers (1%)
- Secondary: All purchasers (3%)
- Tertiary: Engaged non-purchasers (5%)
Value-Based Lookalikes
When available, use value data to weight your seed:
- Include purchase amount or LTV score
- Platform weights users by value
- Lookalike skews toward high-value patterns
Exclusion Strategy
Always exclude existing customers from lookalikes:
- All app users
- Website visitors (if applicable)
- Previous converters
Platform-Specific Tips
Meta (Facebook/Instagram)
- Use Advantage+ lookalikes for automated expansion
- Value-based lookalikes from purchase events
- Minimum 100 users in seed
Google/YouTube
- Similar audiences (being deprecated)
- Move to optimized targeting + first-party data
- Customer Match for seed audiences
TikTok
- Custom and lookalike audiences available
- Pixel-based and customer list options
- App event optimization recommended
Measuring Lookalike Performance
Key Metrics to Track
- Quality: CVR, retention, LTV by lookalike
- Scale: Reach, impressions, spend capacity
- Efficiency: CPA, ROAS compared to other targeting
- Decay: Performance over time as audience saturates
Testing Framework
- Test different seed definitions
- Compare expansion sizes
- A/B test lookalike vs interest targeting
- Measure downstream metrics, not just CPI
Build Better Lookalikes
ClicksFlyer helps you create high-value seed audiences and reach lookalikes across premium inventory.
Get StartedCommon Mistakes
Using all installers as seed
This creates a lookalike of "people who install apps"—not valuable. Use quality signals.
Not refreshing seeds
User behavior evolves. Update your seed audiences quarterly at minimum.
Ignoring platform guidance
Each platform has optimal seed sizes and best practices. Follow their documentation.
Lookalike audiences are a cornerstone of effective mobile UA. Invest in your seed quality, test systematically, and continuously optimize for downstream metrics.