US retail media ad spending is forecast to reach $72.97 billion in 2026, and EMARKETER reported that 73% of advertisers planned to increase their investment in retail media networks.
These networks promise something few advertising channels can: a direct line from seeing an ad to checkout. Retailers can connect shopper data with transactions, giving brands a clearer view of what happened after an ad was seen.
A shopper may see an ad for a brand they already intended to buy and purchase it moments later. The retailer can record the sale. The unresolved question is whether the advertising caused it.
The easiest customers to attribute may be the hardest to move
On-site retail media often reaches shoppers already close to a decision. The people most likely to convert after exposure may also have been the people most likely to convert without it.
A loyal buyer creates an especially attractive attribution record. Purchase history identifies the customer; familiarity makes another purchase more likely; and the transaction can be linked to an impression or a click. The commercial value of the advertising depends on whether it changed the probability, timing, or size of that purchase.
Return on ad spend (ROAS) can be inflated by campaigns that are highly effective at finding likely buyers. A campaign reporting $8 in attributed sales for every dollar spent could generate less additional revenue than one reporting $5 if most of its customers would have purchased anyway. Optimization can reinforce the bias by directing more spend toward shoppers most likely to convert, even when their behavior is hardest to change.
Seeing the sale does not prove what caused it
In November 2025, the Interactive Advertising Bureau (IAB) and IAB Europe issued guidelines for incremental measurement in commerce media. They define incrementality around a causal question: what additional business outcome occurred compared with what would have happened without the marketing activity.
Brands can test this by withholding advertising from a comparable group of shoppers and measuring the difference in purchasing. Where that is not possible, they can compare similar markets in which the campaign did and did not run.
IAB regards controlled tests as stronger evidence that advertising caused an outcome than platform measures without a true comparison group.
What the test measures matters too. Incremental profit can lead to a different investment decision than units sold or new buyers.
“New to brand” does not prove that the advertising created a new customer
For sponsored advertising and demand-side platform (DSP) campaigns, Amazon reviews a shopper’s purchase history with the brand on Amazon over the previous 12 months. If the shopper has not purchased from the brand during that period, a purchase linked to the campaign can be classified as “new to brand.”
That shopper may have bought the brand more than a year earlier, purchased it through another retailer or platform, or formed the intention elsewhere before Amazon recorded the transaction. New-to-brand status describes what Amazon can observe. It does not establish that the advertising created the customer.
Retail media networks need a common test
Networks do not start with the same data. A large retailer with a deep loyalty database can identify shoppers differently from a smaller specialist network. Measurement windows, audience definitions, transaction coverage, and exposure to other media also vary.
A ROAS of six from two networks can describe two quite different economic outcomes.
IAB notes that customer data and measurement systems differ across networks. A test available within one may not work in another, while metrics produced by each platform are difficult to compare directly.
Brands can still apply the same commercial standard. Using a consistent definition of incremental revenue or profit provides a firmer basis for deciding where to allocate investment.
The channel also sits inside a wider media plan. A campaign may amplify preference created by television, video, social media, or another channel. It may reach many of the same consumers and add little. A well-designed experiment can show that the campaign generated additional business without revealing whether the same money would have produced more elsewhere.
IAB identifies marketing mix modeling (MMM) as better suited to comparing investment across the wider marketing portfolio, while experiments provide stronger evidence about individual campaigns. Its newer guidance recommends using both when making budget decisions.
Sales lift still does not tell you what changed
A measured sales increase may recruit a new household, win a buyer from a competitor, or bring forward a purchase that would have happened later. Each has a different commercial value.
Purchase-path research can examine prior brand use, consideration, purchase timing, and retailer choice to identify where the decision shifted.
Results among habitual buyers carry a different implication from those among lapsed customers or people who previously bought a competing brand. Self-reported reasons still require caution. People are imperfect witnesses to their own decisions.
As retailer advertising moves off-site and further up the funnel, exposure happens earlier in the journey, where immediate conversion captures less of the impact advertising may have had. US off-site retail media display spending is forecast to reach $16.09 billion in 2026, compared with $9.63 billion for on-site display. At that distance from checkout, more channels can plausibly influence the eventual transaction.
Advertising may alter consideration without producing an immediate purchase. Longitudinal research and brand tracking can test whether changes in consideration and preference persist beyond the attribution window.
The budget question extends beyond retail media
The channel does not draw from a single pool of money.
EMARKETER reported that 74% of organizations funding commerce media used brand and upper-funnel budgets. Dedicated retail media, ecommerce, and performance budgets are also major funding sources. Nearly half drew from two or three budgets, while almost a third drew from four or more.
Different teams can apply different definitions of success to the same investment. Each can hit its own target while the company allocates more money to activities producing weaker incremental returns. That becomes difficult to see when brand performance, network efficiency, ecommerce conversion, and commercial support are reported separately.
The company needs a single standard for assessing the economic value of the channel, even when different teams use it for different purposes.
McKinsey's 2026 survey of North American grocers found that 64% expected trade funding, retail media, and joint business planning to become mostly or fully integrated over the next two to three years. Only 3% expected them to remain largely separate.
That changes what sits inside a marketing number. Investment in these networks that protect visibility, strengthen the retailer relationship, or support a wider commercial agreement may still be valuable. Calling all of it growth media makes it harder to know whether marketing investment is rising because the company has found a productive source of demand or because the cost of doing business with retailers has changed.
Measure the economics of the incremental sale
An incremental sale can still be a poor investment.
A campaign may generate additional revenue and destroy value once media costs and promotional discounts are factored in. Another may appear modest during the campaign but acquire households that continue buying after the advertising stops.
Additional sales should be translated into profit after media and promotional costs. Newly acquired households can then be followed to determine whether subsequent purchases justify the cost of acquiring them.
Scale matters too: an additional million dollars invested in a network will not necessarily reproduce the return generated by the previous million. Experiments at different spend levels can show where returns begin to weaken and another network or channel becomes the better investment.
The next retail media budget should be built on evidence independent of the network that sold the advertising. Kadence helps brands establish that evidence before the next round of spending is committed.
FAQs
What is retail media incrementality?
Retail media incrementality measures the sales or other business outcomes that happened because of the advertising and would not have happened without it. It goes beyond attribution by asking whether the campaign actually changed customer behavior.
How do you measure incrementality in retail media?
Brands can compare shoppers or markets that received the advertising with comparable groups that did not. The difference in purchasing helps show how much additional business the campaign generated, rather than simply counting sales that occurred after exposure.
Why can retail media ROAS overstate performance?
Return on ad spend can give credit for purchases from shoppers who were already likely to buy. A campaign can report strong attributed sales while generating much less additional revenue once those existing intentions are accounted for.
How should brands compare retail media networks?
Brands should use a consistent definition of incremental revenue or profit rather than comparing each network's reported metrics at face value. Different retailers may use different data and measurement systems, making apparently similar ROAS figures difficult to compare directly.
Does “new to brand” mean a retail media campaign created a new customer?
Not necessarily. Amazon, for example, classifies a purchase as “new to brand” when the shopper has not purchased from that brand on Amazon during the previous 12 months. The shopper may still have bought the brand elsewhere or formed the intention before seeing the advertising.