Retail media works. That's the problem. When a channel produces clean ROAS numbers inside a walled garden, it's easy to keep feeding it, quarter after quarter, without noticing what you're trading away. What you're trading is the customer relationship itself: the email address, the purchase history, the ability to reach that person again without paying a retailer for the privilege.
In brief: Brands that concentrate their growth inside retail media networks, Amazon, Walmart Connect, and their peers, are generating revenue without building an owned audience. Every transaction closed on a retailer's platform belongs to the retailer's data infrastructure, not yours. The ROAS looks efficient in the short term; the economics erode over time because you cannot compound a relationship you don't control. Redirecting even a portion of that spend toward channels that return first-party data changes the long-run math.
The Walled Garden Gives You the Sale, Not the Customer
Owned-audience economics is the compounding return a brand earns when it can reach, re-engage, and learn from its customers directly, without paying a third party for access each time.
The retail media pitch is straightforward: your products are where the buyers already are, the targeting is sharp, and the attribution is tidy. All of that is true. What the pitch omits is that the customer data stays with the retailer. You get a conversion event. They get a profile.
According to BCG, consumers may encounter more than 15 touchpoints before purchasing. Most of those touchpoints happen outside the retailer's walls. Demand is built in one place and captured in another. When capture happens exclusively inside Amazon or Walmart, you're paying to close demand you may have created elsewhere, and you're leaving the resulting customer data on someone else's server.
This is not an argument against retail media. It's an argument against letting retail media become your only growth mechanism, because the moment you need to reach that customer again, you're back at the auction, paying current rates, with no accumulated advantage.
ROAS Is a Snapshot; Lifetime Value Is a Film
The efficiency argument for retail media is real but narrow. ROAS measures what happened in a transaction window. It tells you almost nothing about whether that customer will buy again, what they'll buy, or what it will cost you to reach them next time.
Magnite makes the point directly: CMOs are now being asked to prove incremental sales, not just sales. That pressure is reshaping how brands evaluate media placements, because a channel that produces high ROAS on existing demand while generating zero owned-audience data is doing less work than it appears.
The compounding asset in consumer marketing is the customer list you can reach for near-zero marginal cost. Email, SMS, owned community: these channels don't charge you per impression every time you want to have a conversation. Retail media charges you every single time, and the rates reflect the retailer's leverage, not your brand equity.
Brands optimizing purely for retail media ROAS are, in effect, renting their customer relationships. The rent goes up. The landlord keeps the equity.
Demand Is Portable; Your Infrastructure Should Be Too
Forbes frames this cleanly: demand has become portable. It can be created in one environment, strengthened in another, and captured somewhere else. The implication for brand builders is that the channel where you capture a sale doesn't have to be the channel that owns the customer relationship afterward.
The brands getting this right are running retail media as a demand-capture layer while simultaneously building infrastructure to convert some of that demand into owned relationships. A QR code on packaging. A post-purchase email sequence routed through a DTC touchpoint. A loyalty mechanic that lives on your domain, not the retailer's. None of this is complicated. Most brands just haven't prioritized it because the retail media numbers look fine.
Martech360 notes that brands using retail audiences across CTV, display, and search often produce more activity without necessarily producing more coherent outcomes. The missing piece is usually a first-party data strategy that gives the brand something to carry forward from each campaign, regardless of where the transaction closed.
The comparison worth making is between two growth models:
- Retail media only: High near-term ROAS, clean attribution, zero owned-audience accumulation, full dependency on retailer access and pricing, no compounding.
- Retail media plus owned-audience investment: Slightly lower blended ROAS in the short term, slower to show in quarterly numbers, builds a customer list you can reach directly, creates compounding returns as that list grows and engagement history deepens.
The tradeoff is real. The second model costs more patience. It also builds something the first model never will.
What to Measure When ROAS Isn't Enough
If your marketing team is measured exclusively on retail media ROAS, the incentive structure will keep producing retail media ROAS. That's not a channel problem; it's a measurement problem.
The metrics that reveal owned-audience health are different: email list growth rate, direct-channel repeat purchase rate, customer acquisition cost in owned channels versus paid channels, and the ratio of customers you can reach for free versus customers you have to pay to reach again. None of these show up in a retail media dashboard.
ProDataAI observes that most marketing budgets are still judged on the last click, even as customers meet brands across a far longer and more distributed journey. The last-click frame systematically overvalues channels that sit at the moment of capture and undervalues everything that built the relationship before that moment, including your own channels.
Building measurement that surfaces owned-audience economics alongside retail media efficiency isn't a technical project, it's a prioritization decision. The brands that make it tend to find that their retail media spend looks less efficient in isolation and more rational as part of a system. That's the right frame. A channel that closes sales and returns customer data is doing more work than a channel that only closes sales.
The Compounding Starts When You Own the Next Conversation
The goal isn't to abandon retail media. It's to stop treating it as a complete growth strategy when it's actually a distribution layer. Distribution is valuable. It is not the same as a customer relationship.
Every brand has a version of this decision in front of it: keep optimizing for the metric the retailer reports back, or start building the infrastructure that makes the next campaign cheaper than the last one. The Method blog covers the systems thinking behind that second path in more depth, but the core logic is simple. Owned audiences compound. Rented audiences don't. The brands that figure this out early spend less to grow later, and they do it without asking a retailer's permission.
Start by counting how many of your customers you can reach directly. If that number is small relative to your total customer base, you already know what to fix.
Frequently asked questions
What does it mean to "own" a customer relationship in retail marketing?
Owning a customer relationship means you have a direct channel to reach that person, typically an email address, phone number, or account in your own system, without paying a third party for access. When a sale closes on Amazon or Walmart, the retailer holds that data. You receive a transaction record. The practical consequence is that reaching that customer again requires buying access again, at whatever rate the retailer sets.
Is retail media bad for brand growth?
Retail media is not bad for brand growth. It is efficient at capturing demand near the point of purchase, and its attribution is often cleaner than upper-funnel channels. The problem is concentration: brands that rely on retail media as their primary or only growth channel never accumulate the owned-audience assets that make future marketing cheaper. Used as one layer in a broader system, retail media is a reasonable tool.
How do I build an owned audience if most of my sales happen through retailers?
The most direct approaches are post-purchase touchpoints (packaging inserts, warranty registration, product activation flows) that give customers a reason to connect with your brand directly, and DTC channels that run in parallel to retail, even at lower volume. The goal is not to move all sales off-platform but to convert some portion of retail customers into people you can reach without paying the retailer again.
What metrics should I track to measure owned-audience health?
Track email and SMS list growth rate, direct-channel repeat purchase rate, the ratio of customers reachable for near-zero cost versus customers requiring paid media to reach, and customer acquisition cost in owned channels over time. These metrics don't appear in retail media dashboards, which is part of why they're undertracked. They are, however, the numbers that predict whether your marketing gets more or less expensive as you scale.
Why does owned-audience data compound while retail media spend doesn't?
Owned-audience data compounds because each interaction, a purchase, an email open, a product review, adds to a profile that makes the next communication more relevant and the next campaign more efficient. Retail media spend does not accumulate in the same way: each campaign starts from the retailer's audience, not from a relationship your brand has built. The cost per reach stays tied to auction dynamics rather than declining as your brand equity grows.