New customer acquisition cost is supposed to be a number you can trust on the day you read it. In Q4 it is not, and the size of the gap is the whole story.
October's new customer acquisition cost read $284 in October. By the time holiday shopping ended, those same clicks had settled at $229. A $55 drop, driven entirely by delayed conversions that landed after the click month closed. BFCM clicks fell $40 the same way, from $277 to $236.
Two reasons, and they compound.
First, its attribution window closes before most of those conversions land. The settle happens in the weeks after the window shuts, so the platform never counts it.
Second, it does not verify new versus repeat at the order level. Meta counts a purchase. It does not know whether that purchase grew your customer base or recycled an existing customer. Your nCAC report does, because new versus repeat is checked on every order.
Put those together and you get a number that looks worse than reality on the day you read it, then quietly improves where no platform is looking.
If your blended ROAS looks healthy but new customer growth reads flat, this is usually why. The acquisition is happening. Your measurement just can't attribute it to the click that caused it. That is the difference between deterministic, click-level attribution and a modeled in-platform estimate.
It also reframes the earlier-is-better pattern across Q4. October and Pre-BFCM clicks settle to the lowest new customer costs because they have the most runway for conversions to land before December 31. Early December clicks carry the worst settled nCAC of the quarter at $274, because the year runs out before they finish converting.
Judge new customer campaigns on their settled nCAC, not the in-window reading, and pull the curve by click month so you can see which windows actually built your customer base.
Download the Q4 + BFCM Meta Tips for 2026 report.
We can rebuild the full nCAC curve for every Q4 window from your own order history, with new versus repeat verified at the order level. Book a call and we will walk it with you, backed by our 3x guarantee.
Delayed conversions land in later weeks. October clicks read a $284 nCAC in October and settled at $229 by year end, a $55 drop. BFCM clicks fell $40.
No. Meta counts a purchase but does not verify new versus repeat at the order level, and its window closes before most conversions land, so it cannot produce a settled nCAC curve.
It usually means acquisition is happening but your measurement cannot attribute it to the click that caused it, which is what deterministic click-level nCAC reveals.