Pinterest Was Reporting 0.7x ROAS. The Real Number Was 3.1x — Before the Budget Was Cut.

The situation

A US home decor brand had been running $12K/month on Pinterest for 8 months. Every month, Pinterest Ads showed sub-1x ROAS. The founder was convinced Pinterest was not working and had already scheduled the call to cut the channel. They asked me for a one-hour audit before making the final decision.

The problem

Pinterest's browser tag cannot fire on Shopify's hosted checkout domain. The moment a customer clicked Checkout and moved to checkout.shopify.com, Pinterest lost them completely. With a 14% match rate, only 1 in 7 purchases was being attributed back to Pinterest. The other 6 were recorded in Shopify but invisible to Pinterest's algorithm. The 0.7x ROAS was not a reflection of Pinterest's performance — it was what 14% visibility looks like.

What we did

1. Compared Pinterest attributed purchases against Shopify orders for 30 days — confirmed a 78% gap between real and Pinterest-visible orders
2. Identified root cause: Pinterest tag fired ViewContent and AddToCart correctly, but the Purchase event was blocked by Shopify's cross-domain checkout redirect
3. Implemented Pinterest Conversions API via Shopify's Order Created webhook — purchase events now fire server-side, completely bypassing the checkout domain issue
4. Configured enhanced match with hashed customer email and phone on every event payload
5. Set up deduplication between browser tag and API using unique order IDs

Why it worked

The Pinterest tag appearing healthy in diagnostics was the most misleading part of this audit — it was firing correctly on ViewContent and AddToCart. The problem only showed when you compared attributed purchases to actual Shopify orders. A platform can show green checkmarks on every test while missing 86% of real conversions.