Most eCom brands run Meta ad accounts built for 2018. They stack 15 different interest ad sets, targeting everything from “yoga” to “Lululemon”, hoping to find the perfect buyer. The numbers show why that approach is bleeding cash. Running narrow interest targeting right now inflates acquisition costs by up to 40%. I see this pattern across dozens of ad accounts every quarter.
The conventional advice for beginners tells you to isolate audiences and test them against each other. That advice is wrong. Meta’s machine learning evolved. Granular audience selection restricts the algorithm. Algorithmic broad delivery, guided strictly by your creative, is the only reliable way to scale an eCommerce brand today.
When I was running my own stores, I made the same mistake. I spent thousands of dollars testing “frequent online shoppers” against “engaged shoppers” back in 2019. It worked then. It fails now. My team at Elite Brands deletes those complex setups on day one.
Meta auction mechanics and how to advertise on Facebook today
The Meta auction changed fundamentally over the last three years. Legacy demographic targeting relied on manual inputs. You told Facebook exactly who to look for based on pages they liked or age brackets they fit. That system is dead. Meta now uses a predictive delivery engine. It analyses thousands of real-time data points to find buyers, regardless of what pages they follow.
Audience liquidity is the core