This is the breakdown we almost didn't publish. Not because the results aren't worth sharing — they are — but because the honest account of how we got there involves admitting that the first six weeks were a grind, a rebuild, and a lot of creative that didn't work before we found what did.
The brand: a direct-to-consumer lifestyle accessories label in the UK. Monthly ad budget: £5,000. Monthly revenue at the point of engagement: approximately £18,000. The brief was simple. They wanted better returns from paid. What we found when we got inside their account was both predictable and instructive.
What We Found When We Audited the Account
Their Meta account had accumulated over two years of campaign history and looked, on the surface, like an active, well-managed setup. Fourteen campaigns running simultaneously. Dozens of ad sets with overlapping audiences. A creative library of 60+ assets — most of them either product photography from the website or boosted Instagram posts.
The problems were structural. Audience overlap was cannibalising budget across campaigns, creating internal bidding competition where their own ads were competing against each other in the same auction. The pixel was firing correctly, but the conversion events were misconfigured — they were optimising for Add to Cart instead of Purchase, which taught the algorithm to find people who browse, not people who buy. And the creative, while aesthetically clean, had no hook — every ad opened with the logo and product shot, the two most skippable frames in any video.
ROAS at the point of audit: 1.4×. Industry average for their category: 2.8–3.2×. The gap wasn't a budget problem. It was an architecture and creative problem.
The Rebuild: What We Changed and Why
We reduced the account to three campaigns: one prospecting, one retargeting warm audiences (website visitors and video viewers), one for existing customers (cross-sell and repurchase). We fixed the pixel configuration to optimise for Purchase. We paused all existing creative and started from scratch with a new content brief.
The creative brief was the most important document in this engagement. We identified three audience segments based on purchase data: gifters buying for others, self-purchasers motivated by quality, and aspirational buyers motivated by brand identity. Each segment needed different creative — different hooks, different proof points, different calls to action.
"The account didn't need more budget. It needed a reason to spend the budget it already had more intelligently. Structural clarity came first — results followed automatically."
The Creative Framework That Changed Everything
For prospecting, we built a testing framework with one rule: the first three seconds of every video had to answer the question "why should I keep watching?" for a cold audience who has never heard of the brand. We tested eight hook variations across two ad sets, running each for seven days with equal budgets before eliminating the bottom four and scaling the top two.
The winning hooks were not the ones we predicted. The brand's instinct was to lead with aesthetics — beautiful product footage, premium feel. What actually performed was a direct-to-viewer opening: "If you've ever bought something online that looked nothing like the photos..." followed by the product reveal. Tension first. Resolution second. Product third.
By week four, the prospecting campaign had a 2.6× ROAS — already above their previous account average. By week eight, with three rounds of creative testing and two rounds of audience refinement, it was at 3.8×.
Month Three: Scaling Without Killing Performance
The temptation at this point is always to scale fast — double the budget, double the results. That's not how Meta's algorithm works. Sudden large budget increases reset the learning phase, destabilise delivery, and often cause a temporary ROAS collapse that spooks clients into pulling back just when the account was building momentum.
We scaled by 20% per week, monitoring delivery, frequency, and cost per acquisition daily. We introduced a lookalike audience layer in the prospecting campaign, seeded from purchasers in the last 90 days. We added a post-purchase email sequence to increase repeat purchase rate, which improved the lifetime value calculation and justified more aggressive acquisition spending.
By the end of month three, the account had reached a 4.2× blended ROAS on £5,000 monthly spend, generating £85,000+ in attributed revenue. More importantly, the new customer acquisition cost had dropped by 34% from where we started.
What This Case Study Actually Teaches
The specific tactics matter less than the underlying principles. Fix attribution before you optimise. Reduce complexity before you add budget. Test creative systematically rather than intuitively. Scale incrementally rather than aggressively. Measure new customer acquisition separately from blended ROAS.
Every account we take on goes through the same diagnostic process. The causes of poor performance are almost always structural — architecture, attribution, or creative — rather than budgetary. More spend on a broken account just loses money faster.