Ad prices held steady. The return still fell.
When CPM and CTR are flat, the loss is on your site. Releva works there, and tells your ads who to find.
- Value audiences for Meta, Google and TikTok
- Each order counted once, server-side
- Bigger baskets on your own site
The numbers in Ads Manager do not match the store
Meta claims almost all of what the store made
At a Singapore womenswear brand, Meta-attributed revenue equaled 92% of store revenue from May to mid-August 2026, and 107% in the first half of August.
Some orders count twice, some buyers not at all
At a European bedding and sleepwear brand (bedding, pajamas, blankets and sheets), Meta logged 2.11 purchase events per real order from 4 to 16 July 2026.
You pay to reach your own customers
At a discount sportswear retailer, customers already on the email list clicked 40,545 to 70,579 paid ads in July 2026.
The half of ROAS that lives on your website
ROAS is two things multiplied. Before the click: reach, price and click-through. After it: conversion and items per order.
The agent manages the second half: the products each visitor sees on every page and in every message after the visit.
See the sales agent →Ad tools manage the click. The agent manages what happens after it.
What brands measured
Revenue figures come from store orders.
Shopify revenue per ad click, same 60 days year on year, with value audiences and server-side events.
Read the case study →Same brand, matched 30-day groups+65%Buyer rate of shoppers the value campaign reached: 0.56%, against 0.34% for the old campaign.
Read the case study →Discount sportswear retailer31% lessRevenue credited to email and push in July 2026, against the permissive view-through count. Conservative by design.
Read the use case →Questions from performance marketers
What does Releva do for performance marketers?
It works on both halves of your return. Before the click, audiences built from predicted customer value go to Meta, Google and TikTok. After the click, the agent sells to the visitors you paid for. Each order is counted once, and a holdout group shows the lift in your store data.
How do I keep ROAS when I scale spend?
Give the platforms better signals and stop paying for customers you already have. At a Singapore womenswear brand, the daily Meta budget grew 2.1 times over a 79-day campaign while Meta-reported ROAS held within 8%, with value audiences and server-side events.
How is this different from an attribution tool like Triple Whale?
Attribution tools report what happened. Releva changes what happens after the click and who the platforms look for, then shows the result in your store orders, with a holdout group for the lift.
Our account already double counts. Can we still read results?
Yes. Relative changes against your store's own order count hold up despite double counting. New server-side events share your pixel's event ID, so each order counts once.
Find the half you are not managing
One call with your ad account and your store, side by side.
See it on your store
You'll talk to one of us.