The Invisible Buyer, Part 1

Ad platforms are built for a single advertiser spending their own budget. A good part of retail media fits that model. The brand logs into the network's self-service portal, sets its own budget, and buys directly, and the platform's assumption holds there. A meaningful share works differently. eMarketer projects the US market will pass $109 billion by 2027, more than a quarter of all US digital ad spend. There, the buyer is a retail category manager or shopper-marketing lead. The job is to turn a supplier's co-op funding into advertising: sponsored placements in the retailer's search results, display ads on its site and app, offsite inventory aimed with the retailer's purchase data. They spend money that contractually belongs to a supplier, bound by that supplier's terms, and they reconcile reports from systems that were never designed to agree with each other. This piece follows six people who live it, at six different companies, drawn from five separate pieces of reporting.

Brian Monahan gives the clearest single illustration of why the reconciliation is so hard. Monahan is senior vice president of retail media at Albertsons Media Collective, the retailer's in-house media division. That puts him among the people directly responsible for reconciling these numbers. Asked why two dashboards can show different numbers for the same campaign, he answers with several. In February 2026 he cited an Albertsons white paper finding: 'There are at least 13 variables — attribution windows, UPC inclusion, and more — and those differences can swing ROAS by 30% for the same campaign.' The count comes from the white paper, and Monahan has elsewhere cited a different count (11 variables, a 30–60% swing), so treat the figure as illustrative, not exact. The list is the point.

I write about this ecosystem from the outside. Before ad tech, I spent most of my career on a narrower version of the same question. When two parties both have a legitimate claim on one number, one dollar, one outcome, whose definition of success does the system encode? At Social Bicycles and Superpedestrian, the two parties were a rider and a fleet operator. At Udacity, an employee taking a course wanted the credential's leverage: a title, a case for more pay, a shot at a better project. The company paying for it wanted retention and a workforce it did not have to recruit from scratch. It placed people into courses that matched their level. The employees had chosen differently, for reasons unrelated to the job. In the Merchant ecosystem work I did at Google, the question turned on definitions of trust. Google's view of a quality seller and a marketplace's view of one did not always agree. Retail media is an unusually clean, well-documented version of that question with a third party added: a supplier, a retailer, and the buyer who spends the first party's money on the second party's platform and answers to both. The job title varies by company: category manager, retail media lead, shopper marketer.

The money comes from more than one place

The money comes from three different budgets. A Forrester survey published in January 2025 found that 36 percent of U.S. retail media spend comes from trade marketing budgets and 26 percent from shopper marketing budgets, with the remainder sitting in marketing teams' budgets. The trade-funded share is shrinking: the Path to Purchase Institute's 2025 study found that media budgets grew from 74 percent of retail media spending in 2023 to 80 percent in 2024.

The pools have different owners, and the owners do not always agree. Trade and shopper money generally sits with sales teams; national marketing money sits with marketing and media teams. "Retail budgets are generally sitting with sales teams. National budgets are generally sitting with marketing and media teams, and that is the tension," one CPG marketer told Digiday. Forrester analyst Nikhil Lai drew the line this series follows: "Shopper and trade dollars have strings attached to them that the digital and national dollars don't."

A growing share of brands spend their own money directly. IAB Europe's buying taxonomy divides retail media into three routes: managed service, where the retailer plans and runs the campaign; self-service, where the brand runs it in-house; and programmatic buying via a DSP. Walmart Connect expanded its self-service display offering in 2024, lowering the entry point for brands shut out by managed-service minimums. Purina staffs the function in-house: Nestlé Purina North America employs a Senior Manager of E-Commerce Marketing, Lindsey Hogan, who ran geo-targeted Amazon DSP campaigns through WPP's OpenMind team that delivered a 313 percent higher new-to-brand detail page view rate at 86 percent lower cost per view, work Amazon says positioned the approach as a key component of Purina's 2026 evergreen strategy. Where the brand buys its own media, the platform's single-advertiser assumption holds: one advertiser spending its own money.

But a meaningful share still arrives with someone else's terms attached. Under a joint business plan, a brand commits co-op funding earmarked for specific products and timeframes, and the spend carries reporting obligations back to the brand that paid it. The retailer puts the money to work without owning it. One campaign has to satisfy two budgets with two owners and two sets of terms.

That co-op money then splits two ways. Onsite spend stays on the retailer's own properties: sponsored placements in search results, display ads on the site and app. The retailer's systems tend to know whose money it is, because the budget was committed inside the retailer's own planning. Offsite spend leaves the property. The retailer buys inventory on Google, Meta, Pinterest, and TikTok from its own accounts, aimed with its own purchase data. The platform on the other side sees the retailer as the advertiser. It does not know the dollars began as a brand's co-op commitment, or as trade funds a category buyer is deploying under a joint business plan.

The measurement changes with the move. Onsite, the retailer's attribution at least starts from systems that recorded the money's origin. Offsite, the retailer's attribution has to reconcile with the external platform's own accounting of what its inventory delivered, and that platform never saw the three-party structure at all. That restages the thirteen-variables problem with an additional party at the table. The funding split raises its own measurement question: Lai asked how a brand should evaluate Walmart in a media mix model, counting slotting fees, trade promotion, and fines alongside the media cost, or only the media itself.

Priya Shenoy, director of omni-shopper marketing at Tillamook, described the split to Forbes in February 2025. Trade funds typically cover price promotions and slotting fees. Retail media dollars more often come out of a brand's demand-generation and marketing budgets. Paras Shah, senior director of digital media at Georgia-Pacific, described the same asymmetry from the brand side in an October 2024 Marketing Dive interview. Brands now manage a fragmented list of retailer-run ad platforms. Each is a walled garden built around that retailer's own view of the customer. No single one can stand in for the whole. "That is your full funnel, not my full funnel." The category manager sits where those competing scopes get reconciled into a single report.

Two unrelated pressures compound on top of that split. McKinsey's 2022 research found that more than 80% of spend flowing into retail media networks is incremental to the retailer — including both net-new spend and reinvested brand and performance budgets. That is a claim about where the money comes from relative to the retailer's other revenue, not a claim that the money is net-new to a brand's marketing budget. It is a different claim from incremental sales lift, which is the more contested question in this industry. The distinction matters. Separately, Bain's 2023 survey found only 6% of advertisers said they fully trust retailers' self-reported media metrics. The budget-source problem and the measurement-credibility problem are distinct. They land on the same person.

Inside the retailer, the person who understands the deal often lacks the login

Christine Foster, senior vice president at Kroger Precision Marketing, and Bobby Watts, senior vice president of retail media, digital merchandising, and marketing at Ahold Delhaize USA, describe the same fault line from two different retailers. They spoke in the same October 2025 Modern Retail piece. The person who understands what a supplier needs and the person with the login to the platform running the campaign are often not the same person. The handoff between them is where information and accountability get lost. Foster described the split inside her own organization. Her insights team might work with a brand on one analysis of how its business was doing at Kroger while her retail media team ran an activation disconnected from that analysis. "If the insights team was working with the brand against a certain kind of analysis or viewability into how their business is doing at Kroger, and then we in the retail media side were working with them on an activation that had nothing to do with that analysis, that then was a competing priority internally." Watts called it a structural problem the industry has not solved. "The old days of having just a chief marketing officer, then having a chief commercial officer — or even a merchant team and a marketing team, and then having retail media as this side project — in a retail organization, that's not going to work in the future."

The platforms optimize their own metrics. Nothing in the reporting shows where one supplier's money went.

That gap takes different shapes at different layers of the system. One layer is measurement. The attribution-window and UPC-inclusion disagreements Monahan described live inside a retail media network's own reporting stack. A second layer is automated bidding. Campaign types pool budget across inventory and placements to optimize a single platform-level metric. In 2026, Google's version is called Performance Max and Meta's is Advantage+ Sales (the successor to the former Advantage+ Shopping campaigns). These systems find efficient placements against the metric they optimize. The person accountable for a specific supplier's money gets no view of where that money went, in the language the supplier's contract requires. Milo McMahon, founder and CEO of Outdoor Ecommerce, described the automated-bidding version of this opacity to AdExchanger in 2022. Marketers describing this class of campaign still reach for his language: "It feels totally arbitrary. It's like they're giving us glimpses behind the curtain, but choosing to do so in ways that never tell us the full story." McMahon was not managing supplier co-op money. His complaint is the general one every self-directed advertiser has about black-box automation. It is the same gap in simpler form: a system that reports what it optimized for, and nothing else.

What the six accounts add up to

The piece rests on six people at six companies. They arrived at it independently. The pattern holds anyway. The job requires reconciling systems built for someone else's version of the work. At $109 billion and growing, the scale alone rules out treating this as a niche operational inconvenience. It is what happens when infrastructure gets built around an imagined single advertiser and a meaningful share of the spend underneath it turns out to be multi-party. Foster and Watts state the clearest, most checkable version of it. The person who understands the deal and the person with the login are often not the same person.

Nobody quoted above holds the title that actually originates this money. That title is Category Development Manager, the supplier-side role responsible for co-op budgets. It speaks publicly, when it speaks at all, through adjacent titles: shopper marketing, trade marketing, customer development. Several of those voices are quoted above. Monahan counted thirteen variables behind two dashboards that could not agree. By Part 7, the thirteen will read as one visible instance of a more general gap, showing up in different form at every layer of this job.

References