Most retargeting ROAS is a receipt for a sale you were already going to make. That's not performance — that's bookkeeping.
The retargeting vs prospecting question is the oldest budget debate in commerce media, and most brands get it wrong for a predictable reason: the dashboard flatters retargeting. Users who visited your site, browsed a product, or abandoned a cart convert at multiples of cold traffic, so the retargeting line always looks like your best campaign. Budgets follow the reported number, prospecting gets starved, and six months later the brand wonders why growth has stalled even though ROAS never looked better. The problem isn't the channel. It's what the number is actually measuring.
Retargeting operates on the warmest audience you have — people who already found you, already showed intent, and in many cases were already going to buy. Last-click and platform-reported attribution hand the full sale to the last ad served, which is almost always the retargeting impression that fired minutes before checkout.
That inflates the metric in three specific ways:
If the product was already in the cart, your retargeting ad didn't win the sale. It photobombed it.
Prospecting in commerce media does the harder, more valuable job: putting your product in front of high-intent buyers who haven't met you yet. Its reported ROAS will always look worse, because it's paying for real demand creation instead of collecting credit for demand that already existed.
This is where commerce media earns its keep over generic display. Contextual placement on commerce content, intent-based audiences on Meta, TikTok, and Google, and affiliate publishers reaching in-market readers all target people actively deciding what to buy — just not yet deciding to buy from you. That is the population where a new sale is actually new.
There's no universal ratio, but there are defensible rules of thumb:
The only honest way to settle the retargeting vs prospecting budget question is a holdout test: suppress retargeting for a randomized slice of your warm audience and compare purchase rates. Brands that run this test rarely find zero effect — retargeting usually does something — but the incremental ROAS routinely comes in at a fraction of the reported number, while prospecting's true contribution is understated for the same reason in reverse.
Run the test, reweight the budget, repeat quarterly. Audience dynamics shift, and last quarter's split is not a permanent answer.
The uncomfortable summary: the campaign that looks best in your dashboard is usually the one doing the least new work. Commerce media done properly isn't about harvesting the demand you already own — it's about converting real purchase intent you haven't captured yet into revenue that wouldn't exist otherwise. That's the only ROAS worth scaling.