Custom Audiences in 2025: Setup, Strategy, and iOS Workarounds

Custom audiences were the crown jewel of Meta advertising. Upload your customer list, target website visitors, reach people who engaged with your content. Then iOS 14.5 happened. App Tracking Transparency gutted the signal. Match rates dropped. Retargeting pools shrank. The playbook that worked in 2020 fails in 2025.

This isn't a eulogy for custom audiences—they still work. But the tactics have shifted. Some data sources retained their power. Others became nearly useless. Understanding which is which determines whether your custom audience strategy delivers results or wastes budget on degraded signals.

The iOS 14.5 Impact: What Actually Changed

Before App Tracking Transparency (ATT), Meta's pixel tracked users across websites and apps with near-complete visibility. Website visitors, purchasers, cart abandoners—all captured automatically, all targetable. Then Apple gave users a choice: allow tracking or block it. Roughly 75-80% chose to block.

Signal Loss by Audience Type

Not all custom audiences lost signal equally. The impact varies dramatically by data source:

The pattern is clear: audiences built from on-platform activity (video views, page engagement, lead forms) retained signal. Audiences built from off-platform tracking (website, app) lost significant signal. This distinction shapes everything about custom audience strategy in 2025.

Custom Audience Types That Still Work

1. Customer List Audiences

Customer list uploads remain the most reliable custom audience source. You upload emails, phone numbers, or other identifiers. Meta matches them against their user database. Match rates typically range from 60-75% depending on data quality.

Why they survived: Customer lists don't rely on pixel tracking. You're providing first-party data that Meta matches deterministically. iOS changes don't affect this process.

Best practices for 2025:

• Include multiple identifiers per customer (email + phone + name) to maximize match rates
• Use lowercase emails and remove formatting from phone numbers before upload
• Segment lists by value (high-LTV vs standard customers) for differentiated bidding
• Update lists monthly—stale lists accumulate churn and reduce relevance
• Hash data before upload using SHA-256 for privacy compliance

2. Video View Audiences

Video engagement audiences capture users who watched your video content on Facebook or Instagram. Since this activity happens on-platform, ATT doesn't affect signal quality.

Targeting options:

• 3-second views (broad, high volume)
• 10-second views (moderate intent)
• 25% completion (growing interest)
• 50% completion (strong engagement)
• 75% completion (high intent)
• 95% completion (very high intent, smaller pool)

Strategic tip: Use video view audiences as a proxy for website visitor intent. Create educational or product-focused videos, then retarget viewers with conversion campaigns. This rebuilds the retargeting funnel without relying on degraded pixel data.

3. Page and Profile Engagement Audiences

Users who engaged with your Facebook Page or Instagram profile. Engagement includes likes, comments, shares, saves, profile visits, and message initiations. All tracked on-platform, all retained post-ATT.

Available windows: 30, 60, 90, 180, or 365 days. Shorter windows indicate higher recency and likely higher intent.

4. Lead Form Audiences

Users who opened or submitted your lead generation forms. This is 100% on-platform activity—Meta captures every interaction without any third-party tracking dependency.

Segmentation options:

• Anyone who opened the form (broad awareness)
• People who opened but didn't submit (abandoned leads)
• People who submitted (captured leads)

Abandoned lead audiences are particularly valuable—these users showed intent but didn't convert. Retargeting them with different offers or reduced friction can recover 10-20% of lost leads.

5. Shopping Engagement Audiences

For businesses using Facebook or Instagram Shops, shopping engagement audiences track product views, cart additions, and purchases happening within Meta's ecosystem. Fully on-platform, no signal loss.

Custom Audience Types with Degraded Signal

Website Custom Audiences

Website visitor audiences still exist, but they're incomplete. Meta's pixel now captures only 40-60% of visitors on average—worse on iOS-heavy traffic sources. This means:

• Retargeting pools are smaller than actual visitor volume
• Frequency metrics are artificially low (same user counted multiple times as different sessions)
• Conversion attribution is fragmented

Workaround strategies:

• Conversions API (CAPI): Server-side tracking that sends events directly to Meta, bypassing browser restrictions. Recovers 20-40% of lost signal when properly implemented.
• Broader windows: Use 180-day windows instead of 30-day to accumulate larger pools despite per-day signal loss.
• Event stacking: Target users who completed multiple pixel events (page view AND add to cart) to filter for higher-quality signals.

App Activity Audiences

App-based custom audiences took the biggest hit. With 75-80% of iOS users opting out of tracking, app event data is severely fragmented. Android audiences remain largely intact, but iOS-heavy apps lost most of their targeting capability.

Workaround: Use App Events Optimization (AEO) with broad targeting rather than custom audiences. Let Meta's algorithm find converters instead of trying to retarget a fragmented audience.

Maximizing Match Rates for Customer Lists

Customer list audiences are your most reliable tool. Optimizing match rates directly impacts audience size and targeting effectiveness.

Data Formatting Best Practices

Multi-Identifier Strategy

Including multiple identifiers per customer dramatically improves match rates. A customer with only an email might match at 65%. The same customer with email + phone + name + zip might match at 85%. Meta uses all available identifiers to find users.

Recommended minimum: Email + phone number. This combination typically achieves 70-80% match rates.

Optimal: Email + phone + first name + last name + city + state + zip. Match rates can exceed 85% with complete data.

List Hygiene and Freshness

Stale customer lists degrade performance. Users change email addresses, phone numbers, and Facebook accounts. A list uploaded 6 months ago might have 10-15% lower effective match rate than when first uploaded.

• Re-upload customer lists at least monthly
• Remove bounced emails and disconnected phone numbers before upload
• Segment by recency—customers from the last 90 days will match better than customers from 2 years ago

Building an On-Platform Funnel

The strategic response to signal loss is building audiences within Meta's ecosystem. This means driving engagement activities that Meta tracks directly, then retargeting those audiences.

The Video-to-Conversion Funnel

Stage 1: Run video view campaigns to cold audiences. Use educational, entertaining, or product-showcase content. Optimize for ThruPlays or video completions.

Stage 2: Create custom audiences of 50%+ video viewers. These users showed sustained interest in your content—proxying for the "awareness" stage of the funnel.

Stage 3: Retarget video viewers with conversion campaigns. Since the audience was built on-platform, signal quality is high. CPAs on this audience often beat pixel-based retargeting.

The Engagement-to-Lead Funnel

Stage 1: Run engagement campaigns encouraging likes, comments, and shares. Carousel posts, polls, and interactive content work well.

Stage 2: Build page engagement audiences from users who interacted in the last 30-90 days.

Stage 3: Target engagers with lead generation campaigns using native lead forms. Keep the entire conversion path on-platform.

The Lead Form Retargeting Funnel

Stage 1: Run lead generation campaigns with native forms. Capture emails and phone numbers directly on Meta.

Stage 2: Create an audience of form openers who didn't submit. These are abandoned leads with demonstrated intent.

Stage 3: Retarget abandonments with modified offers—reduced friction, different value proposition, urgency messaging.

Conversions API (CAPI): The Server-Side Fix

Conversions API sends events from your server directly to Meta's server, bypassing browser-based tracking entirely. This recovers a significant portion of lost signal from iOS users.

How CAPI Recovers Signal

When a user converts on your website, the browser pixel might not fire (blocked by ATT). But your server knows the conversion happened. CAPI sends that event to Meta with user identifiers (hashed email, phone, etc.). Meta matches the event to user profiles deterministically.

Signal recovery: Properly implemented CAPI recovers 20-40% of events lost to browser tracking restrictions.

CAPI Implementation Options

• Direct API integration: Your developers send events via Meta's Marketing API. Full control, highest complexity.
• Partner integrations: Shopify, WooCommerce, and major e-commerce platforms have native CAPI support. Enable in settings.
• Gateway solutions: Tools like Stape.io or server-side Google Tag Manager route events through a server container. Moderate complexity.

Deduplication: Avoiding Double-Counting

When running both pixel and CAPI, the same event might be sent twice—once from the browser, once from the server. Meta deduplicates using event_id. Ensure every event sent via both methods includes the same event_id parameter. Without deduplication, you'll see inflated conversion counts and broken attribution.

Custom Audience Strategy by Business Type

E-commerce

• Primary source: Customer list uploads segmented by purchase value (high-LTV, standard, one-time)
• Secondary source: Shopping engagement audiences from Facebook/Instagram Shops
• Workaround: CAPI implementation for pixel signal recovery
• Strategy: Build lookalikes from high-LTV customer lists rather than pixel-based purchaser audiences

Lead Generation

• Primary source: Lead form audiences (submitted, abandoned)
• Secondary source: Video viewers from educational content
• Workaround: Keep the entire funnel on-platform with native lead forms
• Strategy: Use video view campaigns to build audiences, then retarget with lead forms

App Install/Engagement

• Primary source: Customer lists from in-app registration data
• Secondary source: Page engagement audiences
• Workaround: Focus on Android targeting where signal remains strong; use broad targeting for iOS
• Strategy: Build lookalikes from high-value user lists rather than app event audiences

Local Business

• Primary source: Customer lists from POS/CRM data
• Secondary source: Page engagement and review/check-in audiences
• Workaround: Geo-targeted video campaigns to build local engagement audiences
• Strategy: Combine customer lists with location targeting for local retargeting

Common Custom Audience Mistakes

Mistake 1: Relying Solely on Pixel Audiences

Post-ATT, pixel-only strategies underperform. Diversify to customer lists, on-platform engagement, and CAPI supplementation. Pixel audiences should be one input, not the entire strategy.

Mistake 2: Ignoring Match Rate Optimization

Uploading poorly formatted customer lists wastes audience potential. A 55% match rate vs 75% match rate means 27% smaller audience. Clean your data before upload.

Mistake 3: Stale Audience Windows

Using 365-day windows when 30-day windows would suffice. Older audience members have lower intent and relevance. Match window to your sales cycle. Quick-purchase products need shorter windows; considered purchases can use longer windows.

Mistake 4: No Exclusion Strategy

Showing ads to customers who just purchased wastes impressions and annoys users. Exclude recent converters from prospecting campaigns. Exclude customers from acquisition audiences.

Mistake 5: Treating All Custom Audiences Equally

A 95% video viewer is higher intent than a 3-second viewer. A cart abandoner is higher intent than a page viewer. Segment custom audiences by intent level and bid accordingly.

Key Takeaways

• iOS 14.5 degraded pixel-based audiences by 40-55%; on-platform audiences retained 90%+ signal
• Customer list uploads are now your most reliable custom audience source
• Video view and page engagement audiences serve as post-ATT retargeting proxies
• Conversions API recovers 20-40% of lost pixel signal—implement it if you haven't
• Build on-platform funnels: video to conversion, engagement to lead, lead form retargeting
• Multi-identifier customer lists (email + phone + name) achieve 15-20% higher match rates
• Clean, formatted data and monthly list refreshes maintain audience quality

FAQ

Are custom audiences still worth using in 2025?

Yes, but the strategy has shifted. On-platform audiences (video views, engagement, lead forms) and customer list uploads remain highly effective. Pixel-based website audiences still work but require CAPI supplementation to recover lost signal.

What's a good match rate for customer list uploads?

Aim for 65-75% with properly formatted data. Below 60% indicates data quality issues—check formatting, remove invalid entries, and include multiple identifiers. Above 80% is excellent and typically requires complete multi-field data (email + phone + name + address).

Should I use broad targeting instead of custom audiences?

Not exclusively. Broad targeting works well for prospecting when you have strong creative and pixel data. Custom audiences remain superior for retargeting, lookalike seeding, and high-intent segmentation. Use both in a balanced strategy.

How often should I refresh my customer lists?

Monthly for active lists. Quarterly minimum. Customer data decays as people change email addresses, phone numbers, and Facebook accounts. Stale lists have lower effective match rates and reduced relevance.

Does Advantage+ Audience replace custom audiences?

No. Advantage+ Audience uses your targeting inputs as suggestions but expands beyond them. Custom audiences remain valuable for seeding lookalikes, strict retargeting, and providing the algorithm with high-quality signals. They complement Advantage+ rather than being replaced by it.