Best Targeting Options for Facebook App Installs (2025)

In 2024-2025, Facebook's targeting landscape changed dramatically. The detailed interest targeting that defined campaigns for years now consistently underperforms simpler approaches.

Apps that relied on stacking multiple interests to "find their exact audience" are seeing 20-40% higher CPIs than apps using broad, open targeting.

The shift happened because Facebook's algorithm got significantly better at identifying high-value users without manual guidance. When you narrow targeting with specific interests, you're not helping—you're constraining the system from finding patterns you can't see.

Here's what actually works for app install campaigns in 2025.

The Current Targeting Landscape

Facebook offers five core targeting categories for app campaigns, but they don't all deliver equal performance.

1. Broad (Open) Targeting

You select only:

• Location: Country or region
• Age: Broad age range (e.g., 25-54)
• Gender: All, male, or female

Everything else—interests, behaviors, demographics—is left blank.

This gives Facebook's algorithm maximum flexibility to find users who convert, regardless of whether they fit your preconceived assumptions about who your ideal user is.

Current performance: Broad targeting consistently delivers the lowest CPIs across most app categories in 2025. Some agencies report CPI reductions of 84% after switching from interest-based to broad targeting.

2. Lookalike Audiences

You provide Facebook a seed audience (purchasers, subscribers, high-retention users), and Facebook finds people who share characteristics with that seed.

Lookalike audiences come in percentage sizes:

• 1%: Most similar to your seed (smaller audience, higher quality)
• 5%: Moderately similar (larger audience, lower match quality)
• 10%: Loosely similar (largest audience, lowest match quality)

Current performance: 1-3% lookalikes often deliver the best balance of scale and efficiency. They typically perform comparably to broad targeting in CPIs but can show higher downstream conversion rates.

3. Interest Targeting

Target users based on interests, pages they like, or apps they've installed.

Examples: Fitness, Mobile Gaming, Technology Early Adopters, specific competitor apps.

Current performance: Underperforms in 2025. Interest targeting typically increases CPI by 20-40% compared to broad targeting while limiting scale. Only use for extremely niche apps where broad targeting consistently delivers irrelevant users.

4. Behavioral Targeting

Target based on device usage, purchase behavior, travel patterns, or other behavioral signals.

Examples: Android device users, frequent international travelers, digital activity patterns.

Current performance: Rarely provides meaningful improvement over broad targeting. Useful primarily for excluding audiences (e.g., existing users) rather than inclusion targeting.

5. Custom Audiences (Retargeting)

Target users who previously interacted with your app or business:

• Installed your app but haven't purchased
• Visited your website
• Engaged with your Facebook/Instagram content

Current performance: Delivers the lowest CPIs and highest conversion rates but limited scale. Essential for re-engagement but can't drive primary growth.

Recommended Targeting Strategy

Start with these three targeting approaches and compare performance.

Test 1: Broad Targeting

Setup:

• Location: Your core market (e.g., United States)
• Age: 25-54 (or your core demographic range)
• Gender: All (unless your app is gender-specific)
• Everything else: Blank

Budget: $15-20/day for 7-14 days

This is your control test. It establishes baseline performance without targeting constraints.

Test 2: Lookalike Audience (1%)

Setup:

• Source: Your highest-value users (purchasers, subscribers, or top 25% by LTV)
• Size: 1%
• Location: Same as broad test
• Age/Gender: Same as broad test

Budget: $15-20/day for 7-14 days

This tests whether Facebook's lookalike modeling improves user quality compared to letting the algorithm learn directly from campaign performance.

Test 3: Lookalike Audience (3-5%)

Setup:

• Source: Same as 1% lookalike
• Size: 3% or 5%
• Location: Same as other tests
• Age/Gender: Same as other tests

Budget: $15-20/day for 7-14 days

This tests whether expanding the lookalike audience maintains efficiency while increasing scale.

Comparison Metrics

After 7-14 days, compare:

CPI: Which targeting delivers the lowest cost per install?

Install-to-Event Rate: What percentage of installs complete your key value event (purchase, subscription, registration)?

7-Day ROAS: If tracking revenue, which targeting delivers the best return?

Scale Potential: What's the audience size and daily install volume?

Most apps find that broad targeting or 1% lookalikes win on efficiency, while 3-5% lookalikes offer the best scale-to-efficiency balance.

When to Use Each Targeting Type

Use Broad Targeting When:

• You're just starting and have limited user data
• Your app has mass-market appeal (productivity, social, utility apps)
• Previous interest targeting consistently underperformed
• You want maximum scale potential

Expected CPI range: Typically 15-30% lower than interest targeting

Use 1% Lookalikes When:

• You have 500-1,000+ high-value users to build seed audiences from
• User quality matters more than raw install volume
• You're willing to sacrifice some scale for better conversion rates

Expected CPI range: Similar to broad targeting, but higher downstream conversion

Use 3-5% Lookalikes When:

• 1% lookalikes are spending full budget and maintaining efficiency
• You need to scale beyond what 1% audience size allows
• You're seeing audience saturation with narrower targeting

Expected CPI range: 10-20% higher than 1% lookalikes, but significantly more scale

Use Interest Targeting When:

• Your app serves an extremely specific niche (rare)
• Broad targeting consistently delivers irrelevant users who immediately uninstall
• You've tested broad and lookalikes thoroughly and they genuinely don't work

Expected CPI range: 20-40% higher than broad targeting, with limited scale

Use Retargeting Always:

• Run continuous retargeting campaigns at 10-15% of your total budget
• Target app installers who haven't converted, web visitors, and content engagers

Expected CPI range: 30-50% lower than prospecting campaigns

Platform-Specific Considerations

iOS and Android audiences often respond differently to targeting approaches.

iOS Users

• Smaller available audience due to ATT opt-outs reducing Facebook's data
• Lookalike audiences may be less precise due to limited tracking
• Broad targeting often performs better because it maximizes reach

Recommendation: Start with broad targeting for iOS. Only test lookalikes once you have 1,000+ iOS-specific users for seed audiences.

Android Users

• More complete tracking data enables better lookalike modeling
• Larger available audience sizes
• Interest targeting slightly less harmful than on iOS, but still typically underperforms broad

Recommendation: Test both broad and 1% lookalikes. Android lookalikes often show clearer performance advantages than iOS lookalikes.

Targeting Mistakes to Avoid

Stacking Multiple Interests

Adding 5-10 interests doesn't create a "perfect audience"—it creates a tiny audience that prevents Facebook from scaling and slows learning.

If you must use interests, test one interest per ad set, not multiple interests combined.

Targeting Too Narrow Age Ranges

Unless your app genuinely only serves 25-34 year olds, use broader ranges like 25-54. Narrow age ranges fragment your audience and prevent efficient delivery.

Excluding Too Aggressively

Every exclusion you add shrinks your available audience and limits Facebook's optimization flexibility. Only exclude existing users or truly irrelevant segments.

Not Refreshing Lookalike Seeds

Your highest-value users change over time. Refresh lookalike audiences every 30-60 days to incorporate your newest, best users.

The 70/30 Framework

A proven budget allocation across targeting types:

70% of budget: Broad targeting and proven lookalike audiences (prospecting)

20% of budget: Testing new lookalike variations, expanded audiences

10% of budget: Retargeting installers and engagers

This framework ensures you're always testing new approaches without destabilizing your efficient baseline performance.

FAQs

What's the best targeting option for Facebook app installs in 2025?

Broad targeting consistently delivers the best results for most apps in 2025. Set age, gender, and location, then let Facebook's algorithm find your best converters. Recent data shows broad targeting can reduce CPI by up to 84% compared to narrow interest targeting.

Should I use interest targeting for app install campaigns?

Only for highly niche apps. Interest targeting limits Facebook's optimization potential and typically increases CPI by 20-40% compared to broad targeting. Most successful campaigns in 2025 use broad targeting or lookalike audiences.

When should I use lookalike audiences vs broad targeting?

Test both. Start with broad targeting if you have limited data. Once you have 500-1,000 high-value users, create 1% lookalike audiences and compare performance. Many apps find broad targeting scales better while lookalikes deliver slightly higher user quality.

How large should my lookalike audience be?

Start with 1% lookalikes. If they're spending full budget efficiently, test 3% or 5% for additional scale. Avoid lookalikes larger than 5%—they're typically too broad to show meaningful improvement over standard broad targeting.

Should I separate targeting by platform (iOS vs Android)?

Yes. Create separate ad sets or campaigns for iOS and Android. They often respond differently to targeting approaches, and separating them makes optimization and analysis clearer.

---

Targeting in 2025 is about trusting Facebook's algorithm more than your assumptions. Give it good signals through your optimization events, broad targeting parameters, and quality creative—then let it find patterns you can't manually identify.