Criteo Retail Media Run Against Incrementality, Not ROAS

The reported number and the real one: Onsite sponsored products report the sale the retailer can see, which is the sale closest to the checkout. A brand bidding hard on its own name in a category it already leads will read seven or eight times return in the platform and test out under two when held out. We have run that test often enough to say the gap is structural rather than a client failing: the placement is being credited for demand that your trade spend, your shelf position and your last television flight created. So the first thing we establish is what the auction is adding. Where the retailer permits a holdout, whether by store cluster, by region, or by a category-level pulse when no split is available, we run it before the media plan rather than after it, and we size the budget against the incremental figure. Where no holdout is possible, we write that down as a limitation instead of presenting platform return as evidence. Running the account: Bidding sits downstream of availability. A sponsored placement on a line that has gone out of stock at the distribution centre spends the retailer's traffic on a disappointment, and it damages your standing with the category manager long after the campaign ends. We tie bids to the retailer's stock signal where the API exposes it, and to your own supply plan where it does not, so lines under allocation stop bidding before the buyer has to phone about it. Keyword and category targeting do different jobs and are budgeted differently. Category and automatic placements are where a brand takes share inside the consideration set. Exact keyword on your own name is defensive, and it is priced as defence rather than growth. Offsite extensions stay switched off until onsite has a measured contribution, because offsite reporting inherits the same closed-loop attribution and compounds the error rather than diversifying it. We also work the calendar rather than the console. Retail media budget in accounts like yours usually sits inside trade, negotiated in the joint business plan, and the decision that matters is made in that meeting: which promotional periods carry media, what the retailer commits in return, and what the numbers do to the ranging conversation. We will attend that meeting with your national account manager if you want us there, and brief them properly if you do not. Boundaries: We do not resell inventory and we take no rebate from any network, which is why our view on how much you should spend is not paid for by the spending. We do not manage the retailer relationship on your behalf; that is your account team's job and it should stay theirs. We do not produce creative, though we will specify what a placement needs and reject what will not work in it. We will not start an engagement with offsite retargeting. Not for you if: You sell into one retailer, spend under about fifteen thousand US dollars a month with them, and have no category manager contact. At that size the network's own managed service is adequate and cheaper than we are, and we will tell you so. Brands whose product pages on the retailer's site carry no images at the required resolution, no bullet content and no reviews should fix the page first. Media into a bad page is the most reliable way to spend a trade budget and learn nothing from it.