Does retargeting persuade buyers, or follow buyers who were already coming back?

Does retargeting persuade buyers, or follow buyers who were already coming back?

Retargeting consistently posts elite ROAS in attribution reports. The question that ROAS can't answer: are these ads bringing people back, or just appearing in front of people already on their way back?

The experiments. Multiple controlled studies — including a well-known retailer holdout and academic field experiments — randomized site visitors into a group that received retargeting ads and a control group that didn't, then compared return-and-purchase rates. This is the only design that separates retargeting's persuasion from its tendency to chase warm users.

What they found. Results cluster low and sometimes negative. In several documented cases incremental lift from retargeting was in the low single digits despite last-click ROAS of 5-10x; one frequently cited holdout found essentially zero incremental lift — the retargeted and control groups returned at statistically indistinguishable rates. The ads were decorating a journey already in motion.

The nuance. Retargeting is the purest example of attribution bias because, by design, it targets people who already showed intent. It is structurally guaranteed to correlate with conversion and structurally prone to over-claim it. High reported ROAS is almost a symptom of the bias, not a refutation of it.

Bottom line for practitioners: retargeting ROAS is the least trustworthy number in your dashboard — it measures intent you already had, attributed to the ad that followed it. Run a ghost-ad or holdout test before scaling. Some retargeting is incremental; the documented near-zero cases show you cannot assume which until you measure the counterfactual.