[Feature // The Incrementality Problem]
The Advantage+ Trap: Why Meta's algorithmic campaigns cannibalize your existing customers.
Advantage+ made the media buyer's job quiet. One consolidated campaign, minimal audience structure, and a machine that finds the conversion at the lowest predicted cost. The dashboard looks excellent. The P&L often does not.
The mechanism is not mysterious. Meta optimizes for predicted conversion probability, and your existing customers convert at three to six times the rate of cold traffic. Unless you explicitly exclude them, the algorithm learns that the fastest path to a cheap conversion is the customer who bought last month. Prospecting budget gets reallocated to people who were already going to purchase. The ad platform records the revenue, your blended ROAS looks healthy, and new customer acquisition quietly stalls.
We call this the Advantage+ trap because every layer of the reporting reinforces it. The campaign is reported as a single line item. There is no prospecting cell to inspect, no audience breakdown that says “this spend went to people on your list.” The first honest signal usually arrives in Shopify cohort data or the CRM, weeks after the budget has already scaled.
If the platform cannot tell you whether the buyer was already in your database, it cannot tell you whether the media created the customer.
The reporting gap // The buying decision gap
Brand exclusion is a governance control, not a toggle.
The fix begins with a clean, refreshed customer list and a hard suppression rule for every prospecting surface. Upload hashed email and phone data, exclude the list at the ad set level, and keep retention in a separate campaign with its own budget and its own ROAS target. In Advantage+ Shopping campaigns, where exclusion controls are thinner, we typically run a parallel prospecting campaign with an existing-customer budget cap and treat ASC as the retargeting layer, not the growth layer. The point is accounting integrity: every dollar must be attributable to either new or returning demand.
nCAC is the number the board should see.
New customer acquisition cost equals prospecting spend divided by new customers, not total spend divided by total purchases. Once you separate the two, the picture inverts. Accounts with a blended ROAS of 4.7x routinely carry an incremental ROAS near 2.8x and an nCAC 30–50% above the number the platform reports. That is not a tracking error. It is the difference between demand capture and demand creation.
Blended ROAS is a finance convenience. Incremental ROAS is a decision tool. We measure the latter with geo holdouts or a 10% audience suppression cell over 90 days, then reconcile the lift against new-customer revenue. The table opposite summarizes twelve accounts that ran that protocol in Q3.