01The situation
Shopify acquired merchants through a free trial and self-serve signup, with paid media spread across Google, Meta, TikTok, YouTube, Reddit and LinkedIn. The platforms reported plenty of conversions. The question that decided budget was harder: which dollars produced merchants who paid, stayed and sold?
02What I did
- Optimized to paying merchants, not signups.When Shopify rolled out the $1-a-month trial, I moved bidding and reporting to paid conversions. Signup CPA rose about 20% as low-intent signups fell away, and cost per paying merchant dropped about 30%.
- Proved incrementality with geo holdouts.With Data Science I ran geo-holdout tests across paid social. They found $2M+ in spend that wasn't adding merchants, and the budget moved to channels that were.
- Launched TikTok from zero.I took TikTok from a pilot to $800K a year at a CAC 28% below the account average, and made it a core channel.
- Rebuilt Meta after iOS 14.5.Signal loss broke the old structure. I consolidated 20+ ad sets into three to five, moved to the Conversions API, and held CPA within 15% of pre-ATT levels while much of the industry swung 30% to 50%.
- Built the dashboard budget ran on.Ad data joined to merchant outcomes: GMV at 30, 60, 90 and 180 days by campaign. It became the main tool for allocating budget.
- Tested at volume, one variable at a time.40+ structured experiments and 100+ creative variations a quarter, each designed to isolate a single variable so the result meant something.
- Drove app installs with Product and Creative.App install and re-engagement campaigns drove 2.4M+ Shopify App installs.
- Scaled what worked.Media investment grew 4x while holding a 3.2x ROAS, including launches into new markets across EMEA, APAC and LATAM.
03What it moved
- $20M+ a year in paid media managed across six channels.
- 4x scale at a 3.2x ROAS.
- About 30% lower cost per paying merchant on the $1 trial.
- $2M+ in non-incremental spend found and reallocated.
- TikTok from zero to $800K a year at 28% below average CAC.
- 2.4M+ Shopify App installs.
04What I'd tell you
- Platform CPA is where the analysis starts. The merchant who pays is the conversion that counts.
- A holdout test is cheap next to another year of spend on a channel that isn't adding anything.
- Creative volume beats creative opinions, as long as each test isolates one variable.