The 4.5x ROAS That Was Really 2.1x, and What It Cost

Business Overview

Premium Supps is an Australian sports nutrition brand selling direct to consumers on Shopify. The range spans creatine, a methylene blue line and a broader supplement catalogue. Order values are moderate and the category has strong repeat purchase behaviour, which means acquisition cost can be justified over a customer’s lifetime rather than a single order.

The brand was already spending seriously. Roughly 31,600 dollars a month on Meta and over 60,000 dollars a month on Google.

The Challenge

The Facebook dashboard reported a 4.5x return on ad spend. Shopify did not agree. The founder was preparing to increase budget on the strength of numbers that were not describing reality.

This is the most expensive problem in ecommerce media buying, because it does not look like a problem. The account appears to be working, so the natural decision is to put more money into it.

Objectives

• Reconcile Meta reported performance against actual Shopify revenue
• Find and stop spend that was not producing sales
• Rebuild Meta around the audiences that hold up under click attribution
• Scale Google Shopping and Performance Max as the volume engine

Initial Metrics

Audit Findings

Facebook’s default attribution gives itself credit for a sale when someone merely saw an ad without clicking and then purchased anywhere within 24 hours. Those people were often already going to buy. They were existing customers, email subscribers or direct visitors. On this account, 54 percent of reported purchases were view through credit that Shopify never recorded.

Once the account was re read on 7 day click attribution only, the real picture appeared:

• Five ad sets were consuming 18,342 dollars at returns between 0.00x and 1.29x. The largest were March Madness static at 6,072 dollars and 1.08x, a methylene blue static at 6,027 dollars and 1.11x, and a post ID scaling set at 5,175 dollars and 1.29x.
• One methylene blue video had spent 463 dollars at 0.31x, and four methylene blue ads had spent 334 dollars for zero purchases.
• The genuine winner was small. Broad Advantage Plus catalogue remarketing had spent 3,545 dollars at 9.98x real return and an 11 dollar cost per purchase, but it was running at 8.31 frequency, which means it was close to saturating its audience.
• The second best performer was under funded. Broad Advantage Plus top of funnel with exclusions had spent 2,812 dollars at 3.86x and a 19 dollar cost per purchase, at only 4.78 frequency. That gap between spend and available audience was the clearest scaling opportunity in the account.
• Mid tier performers were survivable but not scalable as built. Creatine Only was at 2.00x and a 38 dollar cost per purchase. The 40 dollar plus catalogue test was at 2.27x and 43 dollars.
• One campaign had been named Winner Set. It was returning 1.29x. It had been scaled on inflated attribution, and the name had become the reason nobody questioned it.

Strategy

Stop making decisions on platform reported numbers. Move the account to 7 day click as the decision metric, reconcile to Shopify weekly, cut everything below break even, and move that budget to the one ad set that had audience left to reach.

Then shift volume growth to Google, where the buyer is already searching for the product and the catalogue does the qualifying.

Execution on Meta

• Attribution reset to 7 day click for all decisions, with view through reported separately as context only
• 18,342 dollars of monthly spend paused across five losing ad sets
• Broad Advantage Plus top of funnel scaled first, because 4.78 frequency proved there was audience left to buy
• Catalogue remarketing held at its existing spend rather than scaled, because 8.31 frequency meant extra budget would raise cost without adding sales
• Creative refreshed on the methylene blue line, where static assets had fatigued at frequencies between 3.89 and 5.56
• Creatine Only rebuilt around an offer led angle instead of a product led one

Execution on Google

Google carried the volume. In the May reporting month the account invested 60,826.59 dollars and returned 252,374.78 dollars in primary conversion value, a blended 4.15x. That came from 2,895 conversions at 21.01 dollars each, drawn from 14,081 clicks and 1,407,424 impressions.

The channel split told us where to put the next dollar:

Campaign Structure and Budget Changes

• Shopping was the efficiency engine at 17.10 dollars per conversion and a 21.96 percent conversion rate, so it received budget ahead of Search
• The strongest single unit was a Shopping campaign at 4.45x, 15.14 dollars per conversion and a 22.89 percent conversion rate. Its footprint was small relative to its efficiency, so it was expanded
• Search was capped rather than scaled. At 2.42x and 58.17 dollars of value per conversion it was the weakest active surface in the account
• A dormant Shopping campaign was found holding a 1,500 dollar daily budget while spending 34.89 dollars for the entire month on 510 impressions. That is not harmless. A large unused budget distorts every pacing forecast the account produces
• Four channels were completely inactive: Demand Gen, Display, Video and Smart. Each was assessed on whether it had a job, rather than switched on for coverage

Optimization Process

Weekly reconciliation between platform reported revenue and Shopify revenue became a standing process, not an audit. Every scaling decision on Meta was checked against frequency, because frequency is what tells you whether an audience has room left. Every scaling decision on Google was checked against cost per conversion by channel rather than blended account return.

Results

Key Learnings

Final Business Impact

The brand now runs on numbers that match its bank account. Budget sits behind the two surfaces that actually return, waste has been removed rather than tolerated, and the business is producing close to a quarter of a million dollars a month in paid revenue with a clear view of what is working and why.