Paid media
that pays back.
Powered
by AI.

A U.S. healthcare-certification nonprofit ran a commercial certification brand, a free-education brand, and the parent mission across four paid channels, with no shared measurement layer. Tracking wasn't wired to revenue, so smart bidding could optimize to cheap form-fills instead of profit, and a single blended ROAS number let weak campaigns hide inside healthy account totals.

Built one production tracking container (90 tags, 54 triggers, 66 variables) feeding GA4 and every ad platform under Consent Mode v2, with purchase events as the only primary conversion so bidding optimized to revenue, not cheap leads. Restructured spend into a three-tier ROAS framework (premium, standard, bundle), ring-fenced brand from non-brand, isolated competitor terms, and split Performance Max by audience. Ran the same revenue-fed motion on Microsoft Ads and used Meta as a top-of-funnel free-course engine. Scaled what compounded and cut what didn't: killed a $3.6K display line at zero conversions, paused negative-margin low-geo PMax, and wound down weak entry-tier non-brand, with all three ROAS tiers beating target.

Managed $887K in paid media plus $177K in donated Ad Grants, returning $1.62M in trackable revenue and $855K in commercial gross profit. Held the flagship at 201% to 199% blended ROAS while cost per acquisition improved from $86.91 to $81.00, both under target, through a clean April 2026 handover. The free-education brand ran at 468% ROAS ($12 CPA); Meta drove 40,506 free-course enrollments at $0.51 each.