The Meta Unlock – Halving CAC
TeachMe.to · Growth Marketing Manager · 2026
Summary
Took over TeachMe.to's Meta paid social at an all-time-high CAC and turned it into the company's most efficient acquisition channel — cutting CAC ~50% and growing new customers +22% on 18% less spend across 6 sports and 50+ US metros. Built both the programmatic ad engine and the closed-loop attribution system behind the numbers.
Key Results
Featured Work
UGC ad creative produced with a 25+ creator roster, run across Feed, Stories, and Reels. Reels consistently delivered ~22% below blended CPA.
Fully AI-generated ad creative: coach avatars regenerated from their profile photos, AI venue backgrounds, dynamic star ratings, and city-specific copy. Built and deployed at scale with Claude Code, Python, and the Meta Marketing API.
Approach
Rebuilt the account structure into an ABO/CBO testing framework by sport — Scaling, Geo Expansion, and Creative Testing — across 200+ geo-targeted ad sets, 6 verticals, and 50+ US metros.
Engineered location-intent targeting with 12–15mi venue-radius audiences and Haversine-based coach-supply matching, so ad creative stayed accurate to local coach availability in every market.
Built an automated creative pipeline generating composited Feed (1080×1080) and Stories ads — AI-generated backgrounds, coach-avatar overlays, dynamic star ratings, and venue-specific copy — cutting production from hours to minutes and fueling constant creative testing.
Automated Meta Marketing API operations in Python — campaign builds, creative uploads, audience creation, budget management, and bulk ad-set deploys with rate-limit handling and resume logic.
Rebuilt attribution as a closed loop, joining PostHog first-touch UTM data with Meta spend to replace platform-reported conversions that were undercounting Meta ~3×.
What Was Delivered
Programmatic Meta ad engine (200+ ad sets, 6 sports, 50+ metros)
Automated Feed + Stories creative pipeline
Closed-loop PostHog + BigQuery attribution
Self-serve performance + geo-waste dashboards
Halo-attribution + LTV:CAC model
CEO-reviewed 90-day paid-social strategy
Impact
- Cut Meta CAC ~50% within one quarter and held the gain steady through 2026.
- Grew Meta-driven new customers +22% while cutting spend 18%.
- Corrected a 15–30% attribution miscount and surfaced meaningful spend in zero-return metros for reallocation.
- Diagnosed the #1 conversion leak — ~84% of drop-off at mid-checkout scheduling steps — quantifying a ~2× sales opportunity on flat spend.
Need to scale paid social efficiently?
Let's talk about building a performance engine that cuts CAC while it scales.