The setup: a D2C brand with good ROAS and a ceiling it couldn't break
A direct-to-consumer skincare brand had been running Meta campaigns for 18 months. Their blended ROAS sat at 2.1x — not bad, but their unit economics required 3.0x to be sustainably profitable at scale. Every time they increased budget, ROAS softened. The instinct from their previous agency was to test new creatives and expand audience sizes. They'd been doing this for six months with no improvement.
When we audited the account, the problem was immediately visible. Their top-of-funnel audience was defined by broad interest categories — "skincare," "beauty," "wellness" — plus a lookalike built from all purchasers. The audience was technically correct but had no quality filter. They were reaching 4.2 million people, of whom a small fraction had any meaningful purchase intent. The creative was strong. The audience was diluting it.
What the data said about their actual buyers
We pulled 12 months of purchase data and segmented their customers by LTV, repurchase rate, and acquisition source. Two things stood out. First, customers acquired through their "problem-aware" creative — ads that led with a specific skin concern rather than a product feature — had 2.4x higher LTV than average. Second, customers from specific geographic clusters and age brackets were repurchasing at 3x the rate of their general audience.
Their targeting was built around who might buy once. Their best customers were defined by who would buy repeatedly. Those are very different people, and the campaigns weren't distinguishing between them.
The restructure: narrow the audience, match the message
We rebuilt the audience architecture around three tiers. A high-fit core audience — tightly defined by the behavioral and demographic profile of their top-LTV customers. A lookalike built exclusively from their repeat purchasers (not all purchasers). And a retargeting pool segmented by engagement depth — separating people who had watched 75%+ of a product video from people who had only scrolled past.
Total addressable audience dropped from 4.2 million to 890,000. The team was concerned. We ran the restructured campaigns alongside the original for two weeks to compare. By day 10, the narrowed audiences were outperforming on every metric that mattered. CPM was higher — expected with smaller audiences. But cost per purchase was down 28%, and average order value on first purchases was up 19% because the messaging was specific enough to attract people ready to spend.
The results at 60 days
Blended ROAS moved from 2.1x to 3.4x on the restructured campaigns. Total spend stayed flat. Revenue from paid media increased 34%. More meaningfully, the 90-day repurchase rate on customers from the new campaigns was 31% higher than from the old ones — which meant the ROAS improvement was understated, since it didn't account for downstream LTV.
The insight that drove everything: scale is not about reaching more people. It's about reaching the right people more effectively. When your audience is defined well, creative works harder, budget stretches further, and the customers you acquire stay longer. The D2C brands that scale profitably are almost always the ones that have the discipline to keep their targeting narrow even when the platform is pushing them to go broad.
Results from a real engagement. Specific outcomes vary by category, creative, and market conditions.