How to Fix iOS Attribution Loss on Meta Ads.

In 2021, Apple introduced App Tracking Transparency (ATT) with iOS 14.5. Users could now opt out of being tracked across apps and websites. Most of them did. Meta's pixel — which relies on browser-side data collection — lost visibility into a significant chunk of conversions overnight.

The Scale of the Problem

For most Meta advertisers, iOS attribution loss is between 20% and 40% of total conversions. That means if Meta Ads Manager reports 100 purchases, the true number might be 120–140. The algorithm sees 100 and optimises accordingly — under-investing in audiences and creatives that are actually performing, over-investing in ones that merely appear to be.

The practical effects: ROAS looks lower than it is. CPA looks higher than it is. Campaign decisions made on this data compound the error — you pause a performing ad set because the ROAS dropped, not realising the ROAS only dropped because iOS users aren't being attributed.

What Is Server-Side Attribution (CAPI)?

Meta's Conversions API (CAPI) is a server-to-server connection that sends conversion data directly from your server to Meta's API — bypassing the browser entirely. No pixel. No cookie. No iOS consent gate.

When a user completes a purchase on your site, your server fires an event directly to Meta's API using first-party data you already hold (hashed email, order ID, value). Meta matches this to the user who saw your ad and attributes the conversion correctly — regardless of iOS privacy settings.

How to Implement CAPI: Step-by-Step

1. Create a Meta pixel (if you don't have one). Go to Events Manager in Meta Business Suite → Create a pixel. Install it on your site to capture browser-side events — this still matters as a redundant signal layer.
2. Set up a CAPI gateway. Meta offers a CAPI Gateway (self-hosted) or direct API integration. The easiest implementation for Shopify is through Meta's native integration. For custom platforms, you'll need to fire server-side events using Meta's Marketing API with your system user access token.
3. Deduplicate events. With both pixel and CAPI running, Meta will receive the same event twice. Use the event_id parameter (same value in both pixel and CAPI calls) so Meta deduplicates correctly. Skipping this inflates reported conversions.
4. Pass first-party data fields. The more matching signals you send, the higher Meta's event match quality (EMQ) score. Prioritise: hashed email (em), hashed phone (ph), external ID, client IP address. Use SHA-256 hashing for all PII fields.
5. Verify in Events Manager. After implementation, check Events Manager for event match quality score. Anything above 7/10 is solid. Below 6 means you're missing matching signals — check your hashing and parameter coverage.

What to Expect After CAPI Implementation

Within 7–14 days of properly implemented CAPI, most advertisers see reported conversions increase by 20–40%. This doesn't mean performance improved — it means the true performance that was always happening is now visible to the algorithm. Consequently:

• Reported ROAS increases (your campaigns were actually better than they looked)
• Reported CPA decreases (same cost, more attributed conversions)
• Lookalike audiences improve quality (seeded with more complete conversion data)
• The algorithm re-optimises toward audiences that were silently converting before

How Zephra Handles This Automatically

CAPI implementation is a standard part of every Zephra campaign setup — not an optional add-on. When you connect your ad account, Zephra's Signal Gap audit identifies the gap between pixel-reported and actual conversions. If CAPI isn't implemented, it's the first remediation step before any optimisation begins.

The reasoning is simple: optimising on incomplete attribution data compounds errors. Fixing the data layer first means every AI decision Zephra makes afterward is grounded in the complete picture — not the 60–80% view that pixel-only setups provide.

See how Zephra's Signal Intelligence layer works →  ·  How the 3-step process works →