Rebuilt paid acquisition around server-side tracking and margin-based optimization — so the algorithm chased profit, not cheap clicks.
3.1×
Return on ad spend
-38%
Cost per purchase
+27%
Repeat-buyer rate
The problem
The brand was spending ~$120k/month across Meta and Google and scaling hard, but blended ROAS had slid from 2.6× to 1.7× in two quarters. iOS privacy changes had blinded the pixel, so the algorithm was optimizing for whoever clicked cheapest — coupon-hunters and one-time buyers — while reporting couldn't prove which campaigns actually made money.
Our solution
We treated ads, tracking and margin as one system. Server-side events restored signal, and by feeding real purchase margin and repeat-purchase data back into the platforms, the algorithm started hunting customers who looked like the brand's most profitable buyers — not just the cheapest clicks.
What we did
- Implemented server-side tracking (Conversions API + GA4) to recover post-iOS signal loss and deduplicate against the browser pixel.
- Passed first-party margin and LTV values back as conversion signals so bidding optimized for profit, not raw purchases.
- Restructured account into a lean testing/scaling split and killed 40% of ad sets that looked fine on clicks but lost money on contribution margin.
- Rebuilt the creative testing cadence to 15–20 fresh angles a month, retiring fatigued winners on data instead of gut feel.
- Shipped a live Looker Studio dashboard tying every dollar of spend to contribution margin and a weekly kill/scale report.
The outcome
Within one quarter the account was profitable at a higher spend than where it had been losing money, and the team finally trusted the numbers enough to scale budget with confidence instead of fear.
“For the first time we could scale spend without watching profit evaporate. The dashboard settled every 'is this campaign working' argument in about ten seconds.”