Retail Media Networks: The Third Wave of US Digital Advertising
Amazon and Walmart are forecast to capture 89% of all incremental retail media spending in 2026 — roughly $9.42 billion of the $10.53 billion in net-new investment. There are reportedly 277 retail media networks operating worldwide, and two of them take nearly nine dollars in every ten of the growth. Retail media is the fastest-growing major channel in US digital advertising, and it is also the most concentrated, which changes what “entering retail media” actually means for most brands.
A capability page from Digital, Ecommerce & Performance Marketing in the United States. Retail media is now roughly 30% of all US digital marketing investment, which makes it a core ecommerce marketing channel rather than a trade-budget experiment. See also performance marketing in a consent-required US market. Last reviewed August 2026.
~$71BUS retail media forecast, 2026
#1Fastest-growing channel by percentage
89%Of net-new spend going to two players
75–77%Amazon share of US retail media
~30%Of all US digital ad spend
91.5%Of display bought programmatically
1. Why retail media grew when signal shrank
Retail media is described in current market analysis as the third wave of digital advertising after search and social, and as the most significant new budget category for CPG and retail advertisers. The stated drivers are ecommerce expansion and privacy-driven shifts in data access.
That second driver is the interesting one. As third-party signal degraded across the open web, the value of an environment where the retailer already knows what was bought rose sharply. Brands are reallocating trade marketing and performance budgets toward environments where purchase intent is strongest and attribution is most direct.
Retail media did not win because the creative is better or the audiences are larger. It won because the retailer sees the transaction, and after signal degradation that turned out to be the scarce thing.
2. The concentration problem
The headline growth is real. What it conceals is where the growth lands, and that matters far more to a brand deciding whether to enter.
| Measure | Figure | Source basis |
|---|---|---|
| US retail media, 2025 | $60.32bn | eMarketer H1 2026 |
| US retail media, 2026 forecast | $71.09bn | eMarketer, roughly 18% YoY |
| Net-new spend to Amazon and Walmart | 89% of $10.53bn | eMarketer 2026 |
| Amazon share of US retail media | 75–77% | Analyst estimate, varies by method |
Sources: eMarketer H1 2026 retail media forecast ($60.32bn in 2025, $71.09bn in 2026, roughly 18% YoY, about 30% of US digital ad spend); eMarketer 2026 projection that Amazon and Walmart capture approximately $9.42bn of $10.53bn in net-new retail media investment; analyst estimates of Amazon at 75–77% US share, which vary by methodology. For context, a 2022 Forrester forecast projected $85bn for 2026 — a reminder that scope definitions differ between providers and forward projections should be treated as directional.
The practical reading is that channel size is the wrong number to plan against. Retail media grew from roughly $40 billion in 2022 to a forecast $71 billion in 2026, but if 89% of the growth flows to two networks, the relevant question for most brands is not how big the category is. It is whether you sell through Amazon or Walmart, and what the remaining networks can actually offer you.
3. The networks and what separates them
Major platforms now competing directly with traditional search and social include Amazon Ads, Walmart Connect, Target’s Roundel, Instacart Ads and Kroger Precision Marketing.
| Network | Native strength | Best suited to |
|---|---|---|
| Amazon Ads | Scale and search intent inside the marketplace | Brands already selling on Amazon |
| Walmart Connect | Mass retail reach, in-store plus online | CPG with Walmart distribution |
| Roundel (Target) | Guest data and category curation | Brands aligned to Target’s shopper |
| Instacart Ads | Grocery basket and substitution behaviour | Food, beverage, household |
| Kroger Precision Marketing | Loyalty data depth in grocery | Grocery-led CPG |
Networks named in 2026 US digital ad spend market reporting as competing directly with traditional search and social. Suitability assessments are operational judgement, not vendor claims.
The structural point about all of them
Each network’s advantage is bounded by its own footprint. Closed-loop measurement inside Kroger tells you nothing about what happened at Target. That is not a flaw, but it does mean retail media produces several accurate partial pictures rather than one complete one — which matters when the numbers are aggregated into a single performance report.
4. Trade budget versus performance budget
One of the most consequential details in retail media is where the money comes from. Reporting notes brands reallocating trade marketing as well as performance budgets into these environments.
| Budget source | Traditionally measured on | Consequence in retail media |
|---|---|---|
| Trade marketing | Distribution, shelf, promotion | Now measured on media metrics |
| Performance media | ROAS, CPA | Now measured on retailer-reported sales |
| Shopper marketing | In-store activation | Converging with digital |
| Brand | Awareness | Largely unaffected |
Based on reported reallocation of trade marketing and performance budgets into retail media environments. Interpretation of measurement consequences is operational judgement.
When trade budget becomes media budget, two teams who never shared a metric are suddenly judged on the same dashboard. That is an organisational problem long before it is a channel problem.
5. What closed loop actually means
Closed loop is the central claim, so it is worth being precise about it. These networks offer access to real-time shopper data, conversion tracking and media performance visibility — the advertiser can connect an ad exposure to a purchase inside that retailer.
Illustrative comparison based on documented retail media closed-loop capability. The observation about self-reported measurement is analytical judgement, not an allegation about any specific network.
6. The measurement trap inside the advantage
Retail media’s strength creates a specific risk that is easy to miss because the data quality is genuinely better than the open web.
| Trap | Why it happens | Control |
|---|---|---|
| Seller grades the work | Retailer sells media and reports sales | Independent incrementality tests |
| Harvesting existing demand | Shoppers already in-market | Holdout on branded terms |
| Double counting across networks | Each claims the same shopper | Cross-network reconciliation |
| Cannibalising organic placement | Paying for what you would have won | Test paused-spend periods |
| Partial picture treated as whole | One retailer’s data looks complete | Marketing mix modelling |
| Trade metrics abandoned | Media dashboard is prettier | Keep distribution measures alive |
Risk assessment based on the structural characteristics of closed-loop retail media measurement. Operational judgement.
The most valuable of these is the branded-term holdout. Retail media excels at capturing shoppers already close to purchase, which means a meaningful share of reported conversions may be demand the brand would have captured anyway. Only a deliberate test distinguishes incremental sales from harvested ones.
7. Who retail media suits, and who it does not
| Advertiser type | Fit | Reasoning |
|---|---|---|
| CPG with retail distribution | Strong | Named as the most significant new budget category |
| Brands selling on the marketplace | Strong | Ad and purchase in the same environment |
| D2C with no retail presence | Weak | No transaction for the retailer to observe |
| Services businesses | Poor | Nothing to sell through a retailer |
| B2B | Poor | Wrong buying context entirely |
| Local businesses | Poor | Footprint mismatch |
Fit assessment derived from the mechanics of retail media, which requires the advertiser’s product to be purchasable within the retailer’s environment. Operational judgement.
This is worth stating because retail media is currently discussed as though it applies to everyone. It does not. If a retailer cannot observe your transaction, the entire advantage that justifies the channel disappears — and what remains is display inventory at a premium.
8. The cost trajectory nobody advertises
Every channel that offers superior measurement eventually attracts enough demand to erode its own efficiency. Search did it. Social did it. Retail media is the fastest-growing channel by percentage in a market where ad budget growth is slowing and CPC and CPM pressure is already noted as rising.
The practical instruction is to establish an efficiency baseline on entry and monitor it, rather than assuming today’s returns persist. Early entrants to any measurable channel report unusually good numbers, and those numbers are usually a function of scarcity rather than of the channel’s permanent properties.
9. What this page does not cover
| Not covered | Why |
|---|---|
| Individual network rate cards | Negotiated, confidential and volatile |
| Specific platform bidding mechanics | Change frequently; use vendor documentation |
| Retailer commercial terms | Tied to your supply agreement |
| In-store retail media | Distinct discipline with different measurement |
| Non-US retail media markets | Different networks and dynamics |
Scope statement. Commercial terms in retail media are typically negotiated and confidential; nothing here substitutes for your own network conversations.
10. The 90-day entry plan
Indicative sequencing. The baseline is placed first because without pre-spend sales data there is no way to separate incremental sales from harvested demand.
11. Mistakes to avoid
| Mistake | Why it happens | What it costs |
|---|---|---|
| Taking retailer-reported ROAS at face value | The data looks precise | The seller graded its own work |
| No pre-spend baseline | Eager to start | Incrementality becomes unmeasurable |
| Summing conversions across networks | Looks like total performance | Double counts the same shopper |
| Using it without retail distribution | Channel is fashionable | Premium display with no closed loop |
| Assuming today’s efficiency persists | Early returns are strong | Scarcity mistaken for structure |
| Letting trade metrics lapse | Media dashboard replaces them | Loses sight of distribution and shelf |
Recurring errors in newly measurable channels; illustrative.
12. What changes in 2027
Competition raises the floor price. As the fastest-growing channel in a market where budget growth is slowing, retail media inventory attracts more bidders. Efficiency established at entry should be expected to compress.
Standardisation pressure builds. With multiple networks each reporting on their own footprint using their own definitions, demand for comparable measurement across retailers grows — and advertisers running cross-network reconciliation now will be better placed when it arrives.
Convergence with CTV and in-store. Retail media data increasingly informs targeting beyond the retailer’s own properties, which extends the addressable value but also stretches the closed-loop claim further from its original meaning.
Key Takeaways
- US retail media is forecast near $71 billion in 2026, up from $60.32 billion in 2025 — roughly 30% of all US digital ad spend.
- Amazon and Walmart take 89% of net-new spend. Category size is the wrong number to plan against.
- The advantage is observable purchase, not better creative or bigger audiences.
- The loop closes inside a system that sells the media and reports the result. Independent incrementality testing is not optional.
- Retail media harvests in-market demand well, so branded-term holdouts are the key test of what is genuinely incremental.
- Each network sees only its own footprint. Several accurate partial pictures are not one complete one.
- If a retailer cannot observe your transaction, the channel’s core advantage does not apply to you.
Frequently Asked Questions
How big is US retail media?
eMarketer forecasts roughly $71.09 billion for 2026, up from $60.32 billion in 2025 — about 18% growth and around 30% of all US digital ad spend. Providers define the category differently, so treat any single figure as directional rather than audited.
Why did retail media grow so quickly?
Ecommerce expansion plus privacy-driven shifts in data access. As third-party signal degraded, environments where the retailer already observes the purchase became disproportionately valuable, and brands reallocated trade and performance budgets toward them.
Can I trust retailer-reported ROAS?
Treat it as one input, not as truth. The retailer sells the media and reports the sales, and the measurement covers only its own footprint. Independent incrementality testing and branded-term holdouts are what separate incremental sales from demand you would have captured anyway.
Should a D2C brand with no retail distribution use retail media?
Generally no. The channel’s defining advantage is that the retailer can observe your transaction. Without distribution in that retailer, you are buying display inventory at a premium without the closed loop that justifies the price.
Can I add conversions across different retail media networks?
Not safely. Each network reports on shoppers in its own environment, and the same customer can appear in more than one report. Summing across networks inflates apparent performance and needs reconciliation against total category sales.
Where should the budget come from?
This needs deciding explicitly. Brands are reallocating both trade marketing and performance budgets, which means teams previously judged on distribution and shelf can find themselves measured on media metrics. Agree the source and the success measure before spending.
Will retail media stay this efficient?
Unlikely at current levels. Every measurable channel attracts demand until efficiency compresses, and this is the fastest-growing channel in a market with slowing budget growth and rising CPC and CPM pressure. Establish a baseline at entry so you can see the drift.
What is the single most useful test to run?
A branded-term holdout. Retail media is very good at capturing shoppers already close to purchase, so pausing spend on your own brand terms for a defined period is the cleanest way to see how much of the reported conversion volume was genuinely incremental.
Which network should a brand start with?
Whichever retailer already accounts for meaningful sales volume for you, since that is where a closed loop actually exists and where a baseline is measurable. Starting with the network offering the best commercial terms rather than the best data fit is a common and expensive error.
Conclusion
Retail media is the most important structural change in US performance marketing since social, and the reason is unglamorous: after a decade of degrading signal, someone can finally see whether the ad led to a purchase. That is genuinely valuable and it explains the budget migration better than any pitch deck.
The discipline it requires is to remember who is holding the measuring tape. Each network reports accurately on its own footprint, using its own definitions, while selling the inventory being measured. None of that makes the data wrong. It makes independent testing the price of using it well — and brands that establish a baseline before their first dollar are the only ones who will ever know what the channel actually did for them.
Work With Me
If retail media reports look excellent and total category sales have not moved, that gap is the whole question. It is the first thing I would test.