ABO testing then CBO scaling beat all-CBO by $7 CPA

ABO testing then CBO scaling beat all-CBO by $7 CPA
Pet supplies, 50d, $115k spend, two account structures.

Arm A: pure CBO from day one. Arm B: ABO (Ad Set Budget Optimization, fixed budget per ad set) to find winners, then graduate them into CBO.

CPA:
— All-CBO: $34
— ABO→CBO: $27 ▼$7

In the all-CBO arm, the algorithm starved test ad sets before they had data; half never spent enough to judge. ABO gave each a fair $50/day window first.

Read: CBO is a scaling tool, not a testing tool. Test on fixed budgets, scale on dynamic ones.