Meta CAPI Setup Recovers 87% of iOS-Lost Conversions for Fashion E-commerce Brand

The Problem: iOS 14+ Destroyed Their Attribution

Since Apple's App Tracking Transparency (ATT) update, Meta Pixel accuracy for iOS users has fallen significantly. For this Australian fashion brand, iOS traffic represented 68% of their website visitors — and nearly all of it was going unattributed.

Monthly Meta Ads spend: AU$25,000. Reported ROAS from Meta dashboard: 1.4x. The campaigns were barely breaking even.

But here's what the low ROAS was hiding: Meta's pixel was only seeing 30–40% of the purchases that were actually happening. iOS users were buying — the Shopify orders confirmed it — but Meta wasn't attributing those sales back to the ads.

The result: Meta's algorithm thought the campaigns were underperforming and kept reducing bids on the audiences that were actually converting.

Understanding the Root Cause

When an iOS 14+ user clicks a Meta ad and makes a purchase:

1. Safari's ITP (Intelligent Tracking Prevention) deletes the Meta pixel cookie within 24–72 hours
2. The browser pixel tries to fire at purchase, but the fbclid parameter that links the session back to the ad is gone
3. Meta sees a purchase with no attribution — it can't connect it to a campaign

The browser pixel is fundamentally limited by browser-level privacy restrictions. The only way to recover this attribution is through the Conversions API (CAPI) — a server-to-server connection that sends purchase events directly from your server to Meta, completely bypassing browser restrictions.

Implementation: Meta Pixel + CAPI Dual-Path Setup

Phase 1: Shopify CAPI Integration (Days 1–5)

We implemented CAPI through Stape.io's server-side infrastructure:

• Connected Shopify order webhooks to Stape.io server container
• Configured server-side events: Purchase, InitiateCheckout, AddToCart, ViewContent
• Each server event includes:
  • event_id for deduplication (identical to browser pixel event_id)
  • user_data: hashed email, phone, first name, last name, country
  • custom_data: purchase value, currency, content IDs, content type

Phase 2: Browser Pixel Cleanup (Days 6–10)

Audited the existing browser pixel setup:

• Removed duplicate events firing from both theme code and a third-party app
• Standardised event naming to match server events exactly
• Added event_id to all browser pixel events (UUID generated at session start, stored in cookie)

Phase 3: Deduplication Verification (Days 11–14)

This is the critical step most agencies skip. Without proper deduplication, you end up counting every purchase twice — once from the browser pixel and once from the server.

Deduplication works by matching the event_id parameter. When Meta receives two events with the same event_id (one browser, one server), it counts them as a single conversion.

We verified deduplication by:

• Checking Meta Events Manager for "Deduplicated Events" column
• Confirming server event match rate was above 80%
• Testing purchase flow with Facebook Pixel Helper and network tab simultaneously

Phase 4: Event Match Quality Optimisation (Days 15–21)

Meta's Event Match Quality (EMQ) score determines how well your events are matched to Facebook users. Higher EMQ = more conversions attributed.

We improved EMQ from 4.2 to 7.8 (out of 10) by:

• Adding all available customer fields to CAPI events (email + phone + name + address)
• Ensuring customer data is captured before the purchase event fires
• Using normalised data format (lowercase email, E.164 phone format)

Results at Week 10

The 4.5x ROAS wasn't an improvement from the previous 1.4x — the actual underlying performance hadn't changed dramatically. What changed was that Meta could now see 3x more conversions and retrain Advantage+ to find the right audiences.

What We Learned

1. CAPI doesn't improve your ads — it improves your data. The campaigns were already performing. We just couldn't see it.
2. Deduplication is non-negotiable. Two clients in the same audit period had CAPI set up but without proper deduplication — their reported conversions doubled but actual sales didn't change.
3. Event Match Quality is a multiplier. Every additional user data field you include increases the percentage of events that Meta can match to real accounts.

Related Services

• Meta Ads Management
• Pixel + CAPI Setup
• Universal Tracking