Snap Ads Had 0.8x ROAS on the Dashboard. The Real Number Was 2.7x.

The situation

A Canadian women's fashion brand was spending $14K/month on Snapchat, targeting Gen Z and millennial women aged 18–32. Their Snap Ads Manager showed a persistent 0.8x ROAS — below breakeven. The founder was 2 weeks away from cutting the channel entirely when a Snap account manager suggested they get the tracking independently audited first.

The problem

Their Snapchat Pixel match rate was 9%. Of every 100 purchases driven by Snap ads, only 9 were being attributed back to Snap. The root cause was predictable: 78% of their audience was on iPhone, and iOS App Tracking Transparency meant the browser pixel had almost no ability to identify who was converting. Each event arrived at Snap as an anonymous signal with no user match — statistically useless for attribution. The 0.8x ROAS was not a reflection of Snap's performance. It was what 9% visibility looks like.

What we did

1. Audited Snap Events Manager — confirmed 9% match rate and cross-referenced with Shopify orders for the prior 30 days to quantify the exact gap
2. Deployed a GTM Server-Side container on a custom first-party subdomain, routing all Snap events server-to-server and bypassing iOS browser restrictions entirely
3. Configured the Snapchat Conversions API with hashed customer email and phone on every purchase event, pulled directly from Shopify's Order Created webhook
4. Implemented event deduplication using snap_client_dedup_id to prevent browser pixel and server API from double-counting the same purchase
5. Ran parallel tracking for 14 days, comparing Snap attributed purchases against Shopify daily, before declaring the fix complete

Why it worked

A 9% match rate means Snap's algorithm is making every optimization decision based on a tiny, unrepresentative sample of real buyers. When CAPI sends hashed email and phone directly from Shopify's servers, Snap can match that event to a logged-in Snapchat profile regardless of iOS privacy settings. The 2.7x ROAS was always there — it was just invisible. This is one of the most common situations I see on Snap: the platform isn't the problem, the tracking is.