BLVCK Paris previously ran their own ads, with several highly-specific audiences at low spend and high ROAS. whenever they tried to scale up, ROAS would drop well below their breakeven point.
To resolve that, we first analyzed all existing Facebook pixel events by all possible data breakdowns and learned everything we can about the audience. For instance, gender and age, or product ID. We’ve noticed that the data is skewed by assumptions, so we ran our own campaigns for 2 weeks, to generate pure data. Our initial ROAS barely broke even, but once we ran the numbers, we were able to buy the right combination of placements, demographics, and interests that was both effective and at a lower CPM (the broadest audiences that served our needs).
In addition to cold audiences, we used several lookalike seeds from relevant custom audiences in onion-like formation (excluding audiences from each other in a form of concentric rings to expand audiences with minimal overlap) to allow for scaling in multiple directions at a time: accessories, apparel, technological items, etc.
By Q4 we scaled the account from under $500 a day to as high as $8,000 in a day in weeks of sales while keeping ROAS above the break-even point – generating $300K, $400K, and nearly $500K a month on Shopify alone.