Why retail media changes the calculator's inputs
The default calculator page pictures a business that advertises its own products, takes the order in its own checkout and keeps the retail margin on every sale. Retail media breaks that picture at both ends. The shopper buys in someone else's shop: the retailer bought the stock at a trade price, sets the shelf price and keeps the difference, so what the brand earns from the purchase was fixed when the retailer placed its order. The comparison of wholesale and direct-to-consumer selling places the retailer's margin between those two prices, which is why a retailer sale earns the brand far less than the same product sold from its own store.
Revenue has a different problem. The retailer selling the placements also counts the sales, and the retail media guide describes its reports as the seller's view of its own product: it sets the rule for crediting a sale, the period over which credit is given and whether the brand's other lines count, and the total usually takes in shoppers who would have bought anyway. The FMCG marketing guide values those reports as a brand's rare timely view of its sales at that retailer, but valuable is not causal, so the revenue the calculator is given has to come from a pilot in which placements run in some cells and are withheld from comparable ones.
This is not the marketplace seller's situation either. A business listing its own products on a marketplace keeps the sale and pays fees out of it, which the ROAS calculator page for marketplace sellers handles by taking the fees off the margin. A brand selling wholesale has no sale of its own at the retailer; its share was settled by the trade price before the shopper arrived.
The inputs for retail media
| Calculator input | On the default page | For retail media | Where to find it |
|---|---|---|---|
| Revenue from ads | Attributed revenue for the period | Extra retailer sales measured against held-out regions, store groups or audiences, at the retailer's selling price | The retailer's sales for the test and held-out cells of a pilot it agreed to run |
| Ad spend | Media cost for the period | The package or self-serve spend for the test period, with any data and platform fees | The retailer's invoice or the self-serve billing |
| Gross margin | Everyday gross margin | The brand's margin per unit of the retailer's sales: what the wholesale price leaves after product cost, as a share of the retailer's selling price | Finance margin on the trade price set against the retailer's shelf price |
The margin row is built in two moves. Take the trade price the retailer pays for a unit and subtract what the brand spends to make it and deliver it to the retailer; then divide the remainder by the shelf price, not by the trade price. The second move matters because revenue is counted in retailer sales at shelf prices, and margin and revenue must describe the same unit of money. Where the pilot sold several products, follow the mix actually sold.
The revenue row comes from the pilot's design: what the test cells sold beyond what the held-out cells say they would have sold over the same weeks, the caused revenue of the comparison of iROAS and ROAS. The ROAS calculator page for retargeting campaigns builds that figure from a random audience split; a retail media pilot can withhold placements by region, store group or audience, because the comparison has to be one the retailer agrees to run. Spend is the whole cost of the test period, since packages bundle placements and audiences bought through a demand-side platform can carry their own fees.
Entering the retail media inputs step by step
Agree the rules and the cells before the pilot starts
Ask the retailer for its sale definition, the window it counts over and its rule on the brand's other products, that is, whether a shopper's purchase of those is credited too, and write the answers down. Agree which cells see the placements, which comparable ones do not, and the dates.
Measure the lift from the held-out cells
When the pilot closes, take the retailer's sales of the advertised products in both sets of cells for the same weeks. The extra sales in the test cells are the revenue input; the retailer's attributed total goes in a separate column.
Build the margin on the retailer's selling price
Take the trade price and the brand's cost per unit delivered to the retailer from finance, and divide what the trade price leaves by the shelf price, weighted by the products the pilot sold.
Enter the figures and note both break-even lines
Enter the lift, the full spend and the margin, then write the break-even the calculator returns beside the own-store figure from the break-even calculator, so the gap between the two routes is on the record.
A worked retail media example with illustrative inputs
Sarvoun Pantry reads a supermarket pilot against its holdout
Sarvoun Pantry, an invented cooking-oil brand, sells through two supermarket chains. Before buying the larger chain's retail media package, it asks for a one-month pilot: £10,000 of placements, shown in some of the chain's regions and withheld in matched ones, with the chain's sale definition, window and rule on the brand's other products written down first.
Assume the chain sells a bottle for £5.00, buys it from Sarvoun at a trade price of £2.90, and that making and delivering it costs Sarvoun £1.30; all three prices are assumptions for this illustration. Sarvoun keeps £1.60, which is 32% of each pound of the chain's sales, so its break-even ROAS on retailer sales is 3.13.
At the end of the month the chain reports £41,000 of attributed sales, a return of 4.1. Against the held-out regions, the extra sales come to £26,000. With £10,000 of spend and a 32% margin, that is a ROAS of 2.6 and a gross profit after ad spend of about −£1,680 (£26,000 × 32% − £10,000); the placements needed £31,250 of extra sales to break even. Had Sarvoun entered the chain's £41,000, the calculator would have shown £3,120 of profit: the opposite verdict on the same month.
Sarvoun turns the full package down, writes one reason that could justify a loss, shelf visibility for a new chilli oil the chain will list next season, and asks for a second pilot limited to that product.
ROAS Calculator
Open the ROAS Calculator →Prefilled with the example's inputs: Revenue from ads $26,000 · Ad spend $10,000 · Gross margin 32% → ROAS 2.6× · Break-even ROAS 3.13× · Gross profit after ad spend -$1,680.
The brand and its figures are invented. Type the chain's attributed total into the embedded calculator's revenue box and the profit line changes sign while the break-even stays where it was.
Interpreting ROAS and break-even ROAS for retail media
Start with break-even ROAS, the figure furthest from what the brand is used to. On a margin expressed against the retailer's selling price, the line sits well above the brand's own-store Break-even ROASMinimum ROAS to avoid a loss., and the comparison of retail media and Shopping ads gives the reason: a retailer sale earns the brand its wholesale margin, an own-store sale the retail margin less fulfilment. Keep retail media targets apart from the brand's Shopping targets, or a return that is excellent at home will pass for success here.
Next read ROAS. On these inputs it is the return on the lift, what the placements caused rather than what they were credited with, and the iROAS comparison makes that the figure a budget line is funded on. Record the retailer's attributed ROASRevenue generated per unit of ad spend. beside it with its sale definition, window and rule on the brand's other products, but never enter it or average it in. Followed across pilots, the ratio between the two shows how far the retailer's reporting runs ahead of the incremental effectConversions caused by marketing. its held-out cellsA randomly chosen part of an audience kept from exposure to measure what exposure caused. reveal.
Then read the profit line, which says whether the pilot paid at the brand's real margin. A loss needs a reason stated before the test began, and the retail media guide describes the kind that can carry one: a product the retailer has only just listed, without reviews or an organic ranking yet, may need sponsored placement for a while until it earns that rank. Without that reason on paper, a loss on the lift is simply a loss. Spend divided by margin gives the extra retailer sales the placements had to produce, an easier figure to discuss with the retailer than a ratio.
Read each retailer on its own. The guide to reach and frequency across platforms places retail media networks among the walled gardens: each reports from its own logged-in shoppers and releases no user-level data, so one retailer's results cannot be deduplicated against another's. Two retailers need two runs.
Finally, the calculator cannot see beyond the pilot. The shopper data stays the retailer's, not the brand's First-Party DataData a business collects directly from its own customers, visitors and users., so the verdict says nothing about repeat purchase, and the tutorial on preparing data for a marketing mix model shows how retail media spend joins the model that estimates its longer effect.
Pitfalls when retail media numbers go into the calculator
- Entering the retailer's attributed sales as revenue. The seller's count includes shoppers who would have bought anyway, so ROAS and profit rise without one extra unit sold.
- Using the own-store margin for retailer sales. The retailer keeps the gap between trade and shelf price; a retail margin in the box drags break-even far below the real line and turns losing pilots into apparent successes.
- Blending two retailers into one figure. Their reports cannot be deduplicated and their trade prices are agreed separately, so a combined run hides which retailer is paying for the other.
- Reading a figure before asking for the sale definition, window and rule on the brand's other products. A return built on unrecorded rules cannot be compared with the next pilot.
- Paying for sponsored listings on terms the product already ranks for. Those placements can move sales from the organic listing to the paid one without adding any; check whether total sales hold as that spend tapers.
