What a retail media network is
A retail media network is an advertising business a retailer runs on its own property, selling brands placement in front of its own shoppers and measuring the result against its own purchase data. Kroger Precision Marketing, Walmart Connect, Target Roundel, Amazon Advertising and Costco Velocity are the ones a mid-size CPG brand runs into first, and Tinuiti's 2026 roundup lists more than a dozen beyond them, from CVS Media Exchange to Home Depot's Orange Apron Media (Tinuiti).
The reason a brand-side trade person cares is where the money comes from. The line item usually starts inside the customer investment budget, next to the off-invoice allowances and the feature-and-display funds, because the buyer asking for it is the same buyer negotiating the promo calendar. That makes a retail media network a trade decision wearing a media badge, and it should be held to the same standard: what did it add that would not have happened anyway?
Onsite, offsite, and in-store: the three surfaces
Networks sell three distinct kinds of inventory, and the IAB/MRC Retail Media Measurement Guidelines published in January 2024 cover all three (IAB/MRC via Microsoft Advertising).
| Surface | Where the ad runs | What the brand is buying | How clean the read is |
|---|---|---|---|
| Onsite | The retailer's site and app: search, PDP, category page | Shoppers already in the aisle | Best, and the most pre-qualified |
| Offsite | Social, display, CTV, targeted with retailer audiences | Reach beyond the retailer's traffic | Weaker, exposure and purchase split |
| In-store | Digital screens, cooler doors, endcap panels, audio | Store traffic at the shelf | Weakest, exposure is rarely observed |
Onsite sponsored search is the surface most brands buy first and the one that flatters itself hardest, because a shopper typing your brand name into a retailer's search bar has already made the decision. Offsite extends the retailer's audience onto other properties. In-store is the newest and grew fastest, and it is the hardest to attribute, since nobody records which shoppers walked past the cooler door.
Where the money comes from
Retail media rarely arrives as new budget. It arrives as a reallocation, and the reallocation is quiet because it happens inside a total that does not move. Here is a $3.0M customer investment plan for a mid-size natural brand across two fiscal years:
| Line | FY prior | FY current |
|---|---|---|
| Off-invoice and billback allowances | $2,180,000 | $1,905,000 |
| Feature and display funds | $640,000 | $585,000 |
| Retail media (onsite and offsite) | $180,000 | $510,000 |
| Total customer investment | $3,000,000 | $3,000,000 |
Retail media went from 6% of the plan to 17% without a dollar of new money, funded by $275,000 out of allowances and $55,000 out of feature and display. The scale of that shift across the industry is real: Forrester's forecast has global retail media spending going from $184 billion in 2025 to $312 billion by 2030, and P&G and Diageo have both named media fragmentation as the reason they are moving (Wiss, February 2026).
The practical consequence for a trade planner is that your promoted-volume baseline changes underneath you. Pull $275,000 out of allowances and the promo weeks that money used to fund stop happening, so a year-over-year incremental sales comparison is now measuring two different plans.
Whose number is that?
This is the part worth being blunt about, because the reporting looks authoritative and most of it is self-reported by the party selling the media.
| Metric | Who computes it | Can the brand verify it independently? |
|---|---|---|
| Impressions | The retailer's ad server | No |
| Clicks and CTR | The retailer's ad server | No |
| CPC and CPM | The retailer's ad server | No |
| Viewability | The retailer's ad server | No |
| Attributed sales | The retailer's ad platform, matched to its own loyalty data | No |
| Platform ROAS | Attributed sales over spend, both the platform's numbers | No |
| Units sold at the register | Your POS or syndicated feed | Yes |
Every row above the last one is the retailer's ad-server output. That does not make it dishonest, and the IAB/MRC work exists precisely to standardize the definitions and open them to audit. It does mean the retailer is grading its own exam, over a lookback window it chose, using a halo rule it chose.
The one row a brand can check without asking permission is the last one. Units scanned at the register show up in your retailer POS or syndicated feed whether or not you bought media, which makes incremental units against a baseline the only figure in the stack you can defend to a CFO.
What a brand can actually verify
Run a four-week Kroger Precision Marketing buy for a frozen-snack brand across 1,150 doors and grade it on your own data instead of the dashboard:
| Line | Value |
|---|---|
| Media spend, 4 weeks | $48,000 |
| Doors in the buy | 1,150 |
| Baseline units per store per week (prior 12 non-promoted weeks) | 12.4 |
| Campaign units per store per week | 15.1 |
| Incremental units per store per week | 2.7 |
| Incremental units (2.7 x 1,150 x 4) | 12,420 |
| Brand gross profit per unit | $2.10 |
| Incremental gross profit | $26,082 |
| Return on the media spend | 0.54x |
| Platform-reported ROAS on attributed sales | 6.2x |
Both numbers are arithmetically correct and they answer different questions. The 6.2x counts every shopper who saw an ad and later bought, including the ones who buy the item every second Tuesday. The 0.54x counts only the units that were not there before. On this campaign the brand spent $48,000 to generate $26,082 of gross profit it did not previously have, and no amount of CTR improves that.
Two cautions before you run the same math. Hold trade promotion out of the campaign weeks, or the lift belongs to the deal and the media at the same time. And pick the baseline window before you see the result, because a baseline chosen afterwards is a decision, not a measurement. The measurement playbook for the channel, including holdouts and the gap between attributed and incremental sales, is on incremental sales, and the Walmart Connect specifics are on Walmart's retail media network.
Where Scout fits
Scout reads your retailer POS or syndicated data, models the non-promoted baseline, and reports incremental units and dollars for the weeks a campaign ran, so a retail media network buy can be graded the same way a trade promotion is. To be plain about the boundary: Scout is not an ad platform. It does not buy, serve or target media, and it does not report impressions, CTR, CPC, viewability or the ROAS a network computes. Those numbers belong to the network. Scout's number is lift against a baseline, measured in data you already own.
The short version
- A retail media network is a retailer-run ad business selling onsite, offsite and in-store placement against the retailer's own shopper data. Kroger Precision Marketing, Walmart Connect, Target Roundel and Amazon Advertising are the ones most brands meet first.
- It is usually funded out of the customer investment budget rather than new money, which quietly changes the promoted-volume baseline you compare against next year.
- Impressions, CTR, CPC, viewability and platform ROAS are the retailer's ad-server numbers. The only figure a brand can verify on its own is incremental units against a baseline.
- In the worked example the network reported 6.2x and the POS said 0.54x on 12,420 incremental units. Grade the buy on the second one.