The Challenge
A mid-sized e-commerce brand had been running Meta Ads profitably for two years before iOS 14 broke their attribution model. In the 18 months following the iOS changes, ROAS had declined steadily from a peak of 3.1x to under 1x. The team had responded by increasing spend, testing more creatives, and switching campaign objectives repeatedly, none of which stabilised performance.
The underlying problem was that the brand's entire optimisation strategy had been built on pixel-reported ROAS, which was now severely underreporting conversions. Meta's algorithm was making budget and audience decisions based on phantom data. The creatives that appeared to be performing were often not the ones driving real revenue, and the audiences Meta was targeting had drifted significantly from the brand's actual buyers.
The Approach
The first priority was fixing the measurement layer. We implemented server-side conversion tracking via the Conversions API, passing purchase events with order value directly from the backend to Meta. This restored roughly 60% of the conversion signal that had been lost to iOS attribution gaps. We also implemented a simple blended ROAS calculation using Shopify revenue data versus Meta spend to give the team a reliable performance view independent of platform reporting.
With clean data in place, we audited the creative library. Three years of ads had accumulated with no systematic organisation by message angle, format, or funnel stage. We categorised every active creative into one of four angles: problem-aware (pain), solution-aware (product benefit), social proof (reviews, UGC), and direct response (offer-led). Performance by angle across the restored conversion data revealed that social proof creatives, particularly unedited customer video reviews, significantly outperformed all other angles.
We restructured the account to separate prospecting and retargeting clearly, with Advantage+ Shopping Campaigns for prospecting and a manual retargeting campaign for warm audiences. Creative rotation was put on a structured 14-day testing cycle, with new angles tested against the control in a controlled split rather than the previous approach of launching multiple new creatives simultaneously with no control variable.
The Results
ROAS recovered from 0.9x to 3.4x over 10 weeks. The improvement was driven by three compounding factors: restored conversion signal giving the algorithm accurate data, budget reallocated toward the social proof creative angle that was demonstrably outperforming, and the Advantage+ prospecting campaign finding new buyers at significantly lower CPM than the previous manual targeting setup. CPA dropped 44% as Meta's algorithm, now working with real purchase data, optimised away from low-quality clicks. The 14-day creative testing cycle gave the brand a repeatable process for scaling winning creatives and retiring underperformers before they degraded account performance.
Let's build your growth plan
I'll audit your current setup — ads, tracking, and funnel — and come back with a clear, prioritised roadmap. No fluff, no retainer commitment.
Book a free audit
Digital marketing consultant — SEO, PPC, analytics & CRO.
