Attribution — What iOS14 Broke and How to Work Around It

11 min · text · Advanced

iOS14 broke attribution in 2021 and we have spent four years patching it. Meta reports 15-30% lower conversions than actually happened. Shopify reports different numbers than Meta. You cannot trust any of them fully. Today we build a truth layer.

What broke and why

In September 2021, Apple released iOS14.5, which required apps to ask permission before tracking users across websites. Most iOS users said "no." This meant:

Meta pixel can no longer fire reliably on iOS devices
Last-click attribution became incomplete (iOS orders are often invisible to Meta's pixel)
Meta's Aggregated Event Measurement (AEM) was Meta's band-aid: they estimate iOS conversions based on pattern matching from the Android cohort

The result: Meta reports 0,000 in conversions on a day when you actually sold 3,200. The difference is iOS orders that went invisible.

Shopify's attribution is different from Meta's. Shopify can see orders (they process the payment), but Shopify's attribution relies on the Meta pixel firing correctly. If the pixel doesn't fire, Shopify still records the order but cannot always connect it to the original Meta ad.

Three layers of truth

You need three data sources:

Layer 1: Shopify orders — the ground truth. Shopify sees every order because it processes the payment. 3,200 in orders means 3,200 in orders, full stop. Use this as your revenue baseline.

Layer 2: Shopify attribution — which Shopify orders came from Meta, organic, direct, email, etc. This is 70-85% accurate for Meta orders (the rest are dark traffic — direct, dark social, iOS safari redirect). Bias: understates Meta contribution on iOS.

Layer 3: Meta's reported conversions — what Meta claims you got. This is typically 50-70% of actual Meta-attributed Shopify orders due to iOS gap and pixel fire failures. Use this for daily campaign decisions, but trust the Shopify baseline for weekly P&L.

Rebuild your data layer

Stop using Meta's "Return on Ad Spend" as gospel. Yes, Meta will tell you a $500 campaign produced ,400 in conversions. What Meta means is " ,400 in conversions we detected via pixel fire." The real number is likely ,800-2,000 on iOS-heavy audiences (AU, UK).

Use Shopify's "attributed to Paid Search/Social" report as your second opinion. Shopify sees the order, then works backward to figure out which channel referred it. For Meta orders, this is more accurate than Meta's pixel, but still has a 15-25% iOS gap.

Calculate blended ROAS using Shopify revenue, not Meta revenue. At the end of week 1, tally:

- Total ad spend (from Meta ad account): $3,200 - Total Shopify revenue from orders tagged "Facebook Ads" or "Paid Social": $7,100 - True ROAS: 2.22 (not what Meta claims: 2.65)

Sync daily Shopify revenue to a private spreadsheet. Plot it against daily ad spend. This is your source of truth, not Meta's dashboard.

UTM-tagged links and post-purchase surveys (the patch layer)

iOS attribution will never be perfect. But you can add two repair layers:

UTM parameters: Every Meta campaign should have a UTM that identifies the audience, creative, and campaign. Example: `` ?utm_source=meta&utm_medium=paid_social&utm_campaign=kitchen-gadgets-jan&utm_content=creative-A ``

Shopify's admin reports can filter orders by UTM, giving you a second attribution channel. Bias: only works if the customer visits your store directly from the ad (not true for iOS app-to-web redirects, estimated 10-20% of iOS traffic).

Post-purchase surveys: Add a lightweight post-purchase survey ("How did you hear about us?") to your Shopify thank-you page or post-purchase email. Three options:

Meta / Facebook / Instagram
Google / Search
Other (TikTok, referral, friend, etc.)

A 10-15% response rate gives you an additional 500-1,000 data points per month. Bias: self-reported, but better than nothing.

The weekly (not daily) ritual

Daily: use Meta's reported conversions for campaign decisions — pause underperformers, scale winners.

Weekly (Friday): compare Meta's claimed conversions to Shopify's actual orders. If the gap is wider than historical trend (usually 15-30% under-reported), something changed:

iOS traffic increased
Pixel is not firing correctly
Audience quality degraded

Investigate once a month. Do not obsess.

iOS rate by market

Australia: ~65% iOS, ~35% Android. Meta's AEM underreports by 20-25%. Shopify attribution is typically the more reliable number for weekly P&L.

United States: ~55% iOS, ~45% Android. Meta's AEM underreports by 18-22%.

United Kingdom: ~67% iOS, ~33% Android. Meta's AEM underreports by 20-26%.

The higher the iOS rate, the worse Meta's attribution. Markets with high iOS adoption (UK, AU) are worse than the US for this reason.

Why this matters

Operators who trust Meta's reported conversions often scale losers. The ROAS looks great (2.8) but the true ROAS (after iOS gap) is 1.8, and you are losing money. Operators who compare Meta to Shopify weekly catch this drift and adjust. The truth is in Shopify. Everything else is an estimate that feels good when it should not.

Meta claims 2.8 ROAS, Shopify tells a different story

An operator running a beauty gadget in week 3 of 2026:

Meta dashboard reports:

Ad spend: