Retail media networks have moved from an Amazon curiosity to one of the fastest growing categories in all of advertising, and in 2026 they sit at the center of how brands plan their digital budgets. A retail media network is the advertising business a retailer builds on top of its own store, website, app and first-party shopper data, selling sponsored placements and audience targeting to the brands that already sell through it. What began as Amazon monetizing search results inside its marketplace has spread to Walmart, Target, Kroger, Instacart, DoorDash, Costco and dozens of grocers, drugstores and specialty chains. For retail and e-commerce teams, understanding how these networks work is no longer optional, because a meaningful share of trade and shopper marketing dollars now flows through them.
This guide explains what retail media networks are, how the auctions and data actually function, where the inventory lives across on-site, off-site and in-store surfaces, and how to build a practical plan that does not waste budget. It sits inside the Marketing cluster on ShopAppy and pairs with our wider view of how the channel mix is shifting in our guide to retail marketing in the age of AI search and social commerce. The focus throughout is the US market, with real examples from the retailers driving the category.
In short
- A retail media network sells advertising against a retailer’s own first-party shopper data and shopping surfaces, closing the loop between an ad impression and a verified purchase in a way that open-web advertising cannot.
- Amazon Ads remains the anchor of the category at well over $50 billion in annual ad revenue, but Walmart Connect, Target Roundel, Kroger Precision Marketing, Instacart and a long tail of grocers and retailers now compete hard for the same brand budgets.
- The money is shifting from traditional trade promotion and open-web display into retail media because the measurement is cleaner: closed-loop attribution ties spend to sales of specific products at specific stores.
- The most common mistakes are over-indexing on branded search defense, ignoring incrementality, treating each network as identical, and failing to negotiate data access and reporting transparency up front.
- A workable 2026 plan starts with two or three networks, separates always-on lower-funnel coverage from incremental upper-funnel tests, and insists on third-party verification before scaling spend.
Why retail media networks matter in 2026
Three forces have pushed retail media from a niche line item to a board-level priority. The first is the slow death of third-party cookies and the rise of privacy rules that make open-web targeting less reliable. Retailers sit on logged-in, deterministic purchase data that does not depend on cookies, which makes their audiences uniquely durable. That data advantage is the single biggest reason brand budgets keep migrating toward retail media.
The second force is measurement. A retailer can connect an ad impression to a basket containing the advertised product, which produces a return on ad spend figure that traditional brand advertising has always struggled to match. For a packaged goods brand or a direct-to-consumer seller, that closed loop turns marketing from a cost center into something closer to a sales channel. The clarity is seductive, even when the underlying attribution deserves more scrutiny than it usually gets.
The third force is margin. Retail is a low-margin business, and advertising revenue carries margins that can exceed 70 percent or more. For a grocer running on net margins of 2 to 4 percent, a high-margin media business materially changes the profit picture, which is why nearly every large US retailer has launched or expanded a network. The same dynamic that pushed Amazon to build the category now pulls every competitor into it.
The result is a market that industry analysts size in the tens of billions of dollars in the United States alone, growing faster than search and social combined. For context on how large the broader advertising and retail economy is, the US Census Bureau retail trade data shows total retail sales running well above $7 trillion annually, and retail media is steadily capturing a slice of the marketing budgets attached to that spend. The momentum mirrors the broader move toward measurable, in-environment advertising that we cover in our analysis of in-store retail media moving from pilot to scale.
What is a retail media network? Key terms and definitions
At its simplest, a retail media network is a retailer acting as a media owner. It packages its digital and physical real estate, layers on its shopper data, and sells access to brands that want to reach those shoppers at or near the point of purchase. The retailer controls the inventory, the targeting and the reporting, which gives it a structural advantage over independent ad networks.
Because the vocabulary trips up many teams new to the channel, it helps to fix the core terms before going deeper. The definitions below are the ones you will hear in vendor pitches and quarterly business reviews.
The terms you actually need
- First-party data: shopper information the retailer collects directly through loyalty programs, accounts and purchase history. It is the asset the whole model is built on.
- Closed-loop attribution: matching an ad exposure to a real purchase by the same shopper, online or in store, so the retailer can report sales driven by the campaign.
- On-site media: sponsored products, sponsored brands and display placements that appear on the retailer’s own website and app.
- Off-site media: ads the retailer serves on other properties, such as social platforms, connected TV or the open web, using its shopper audiences for targeting.
- In-store media: digital screens, audio, sampling and shelf placements inside physical stores, increasingly tied back to the same data spine.
- ROAS: return on ad spend, the revenue attributed to a campaign divided by its cost. The headline metric, and the one most often misread.
- Incrementality: the additional sales a campaign generated that would not have happened anyway. The metric that actually matters, and the hardest to measure.
- Clean room: a privacy-safe environment where a brand and a retailer can match data sets without either side seeing the other’s raw records.
The gap between ROAS and incrementality is where most retail media debates live. A high ROAS on branded search often captures sales that would have happened regardless, which means the reported return overstates the true contribution. Sophisticated teams treat ROAS as a directional signal and push for incrementality testing before they trust the number.
How retail media networks work in practice
Most on-site retail media runs through an auction, much like search advertising. A brand bids to have its product appear in a sponsored slot when a shopper searches a keyword or browses a category, and the highest effective bid that meets relevance thresholds wins the placement. The shopper sees a labeled sponsored result, clicks or buys, and the retailer records the outcome against the campaign.
The data layer is what separates retail media from generic search advertising. Because the shopper is usually logged in, the retailer knows what that person has bought before, how often they shop the category, and whether they converted after seeing the ad. That lets brands target audiences such as lapsed buyers of a competing product or households that buy the category but not the brand. The targeting precision is the product, and the inventory is just where it gets delivered.
Off-site, the retailer extends the same audiences to external inventory. It might serve a connected TV ad to households that recently bought a competing brand, then measure whether those households purchased the advertised product in store over the following weeks. This stitching of exposure to outcome across channels is the capability brands cannot replicate on their own, and it is why retail media spend keeps consolidating around the largest data owners.
The role of self-serve platforms and managed service
Networks typically offer two ways to buy. Self-serve platforms let a brand or its agency manage campaigns directly, adjusting bids, budgets and creative in close to real time. Managed service hands execution to the retailer’s team, which suits brands without in-house retail media expertise but reduces control and transparency.
Larger advertisers increasingly demand application programming interface access so they can pull performance data into their own dashboards and bid algorithms. The maturity of a network’s self-serve tooling has become a real differentiator, and it is one of the first things a brand should evaluate before committing budget. Amazon and Walmart lead here, while many smaller networks still lean heavily on managed service.
Amazon, Walmart, and beyond: the US retail media landscape
Amazon Ads is the gravitational center of the category. It pioneered sponsored products inside marketplace search, scaled into display and video through its owned properties, and built an ad business generating well over $50 billion a year. Its scale, shopper intent data and Prime ecosystem make it the default first network for most consumer brands, and its reporting depth sets the bar competitors chase.
Walmart Connect is the most credible challenger. With the largest physical retail footprint in the United States and a fast growing e-commerce business, Walmart can offer brands a blend of online and in-store reach that Amazon cannot match on the physical side. Its acquisition history in advertising technology and its growing connected TV reach through Vizio give it ambitions well beyond on-site search.
Beyond the two giants, the landscape is crowded and specialized. The table below maps the major US networks and what distinguishes each.
| Network | Parent retailer | Core strength | Best fit for |
|---|---|---|---|
| Amazon Ads | Amazon | Scale, search intent, mature self-serve and reporting | Almost every consumer brand |
| Walmart Connect | Walmart | Online plus in-store reach, CTV via Vizio | CPG, grocery, general merchandise |
| Roundel | Target | Design-led audiences, guest loyalty data | Style, beauty, home brands |
| Kroger Precision Marketing | Kroger | Deep grocery loyalty data, household panels | Food, beverage, household CPG |
| Instacart Ads | Instacart | Cross-retailer grocery reach, basket data | Grocery and CPG seeking breadth |
| Sam’s Club MAP | Walmart | Membership data, high purchase frequency | Bulk and club-channel brands |
The grocery networks deserve particular attention because food and household goods generate frequent, predictable purchases that produce rich data. Kroger Precision Marketing, built on the company’s long-running loyalty analytics, is widely regarded as one of the most data-mature networks despite being far smaller than Amazon. Instacart, meanwhile, sells advertising across many grocers at once, giving CPG brands a way to reach the grocery basket without negotiating with each chain separately, a model we examined in our look at how Instacart is being re-rated as grocery-tech infrastructure.
Beyond traditional retail: delivery, travel and connected TV
The category is spilling past conventional retailers. DoorDash and Uber Eats now run advertising businesses against their delivery data, airlines and hotels are launching media networks on their booking and loyalty platforms, and financial institutions are exploring commerce media on transaction data. The common thread is first-party data attached to a purchase or booking moment.
Connected TV is the fastest growing extension. As retailers tie their shopper data to streaming inventory, brands can run television-style advertising with retail-grade measurement, a combination that was impossible a few years ago. The Fox and Roku tie-up is one signal of how shoppable television and retail data are converging, a shift we unpacked in our coverage of the $22bn deal bringing shoppable-TV commerce to scale.
On-site, off-site, and in-store: where the inventory lives
Retail media inventory now spans three environments, and treating them as one undifferentiated pool is a recipe for wasted budget. Each surface has a distinct role in the funnel, a distinct cost profile and a distinct measurement reliability.
On-site inventory is the closest to the purchase and the easiest to measure. Sponsored products that appear in search results convert well because the shopper is already in buying mode, which is why this is where most brands start. The risk is that on-site search can become an expensive defense of sales the brand would have won anyway.
Off-site inventory reaches shoppers earlier, when they are browsing social platforms or watching streaming content rather than actively shopping. It is harder to measure cleanly because the conversion happens later and across channels, but it is where genuine demand creation tends to live. In-store inventory, from digital screens to audio and sampling, closes the gap at the physical shelf and is increasingly tied back to the same loyalty data.
| Surface | Funnel role | Measurement reliability | Typical risk |
|---|---|---|---|
| On-site search | Lower funnel, conversion | High, direct attribution | Paying for organic sales |
| On-site display | Mid funnel, consideration | Moderate | Low viewability, banner blindness |
| Off-site social and CTV | Upper funnel, demand creation | Lower, multi-touch | Weak incrementality proof |
| In-store screens and audio | Point of purchase | Improving, loyalty-linked | Hard to isolate effect |
The practical takeaway is to match the surface to the objective. Use on-site search to capture intent and protect shelf position, use off-site to build demand among lapsed and competitive buyers, and use in-store to influence the final decision. Buying every surface from a single network because it is convenient usually means overpaying for the easy conversions and underinvesting where incremental growth actually comes from.
Measurement, attribution, and the clean-room question
Measurement is where retail media either earns its budget or quietly wastes it. The headline ROAS that networks report is calculated using their own attribution rules, which typically credit any sale within a window after an ad exposure. That self-grading is the central tension of the category, because the party selling the media also marks its own homework.
The fix is incrementality testing, usually through holdout groups where a portion of the target audience is deliberately not exposed to the ad. Comparing the exposed and unexposed groups reveals how many sales the campaign actually caused rather than merely observed. Networks that offer robust, transparent incrementality measurement should be trusted with more budget than those that only report attributed ROAS.
Clean rooms have become the standard mechanism for brands to verify performance against their own data. A brand can match its customer file with the retailer’s exposure data inside a privacy-safe environment, then analyze incrementality without either party exposing raw records. Demanding clean-room access and third-party verification is one of the clearest markers of a mature retail media program.
Standardization is coming, slowly
Industry bodies have published measurement standards to make networks comparable, covering definitions for impressions, viewability and attribution windows. Adoption is uneven, and brands still face a patchwork of metrics that resist apples-to-apples comparison. Until standardization is universal, the burden sits with the advertiser to normalize reporting across networks before drawing conclusions.
Common mistakes and how to avoid them
The same errors recur across brands new to retail media, and most of them stem from treating the channel as free incremental sales rather than a budget that demands the same rigor as any other. Avoiding them is less about clever tactics and more about discipline.
The first mistake is over-investing in branded search defense. Bidding on your own brand terms produces gorgeous ROAS because those shoppers were already looking for you, but much of that spend simply taxes sales you would have made for free. A holdout test usually reveals that a large share of branded search spend is not incremental, and that budget is better redeployed against competitive and category terms.
The second mistake is treating every network as interchangeable. Amazon, Walmart and Kroger have different audiences, data depth, auction dynamics and reporting, and a strategy copied from one rarely performs identically on another. The third mistake is ignoring incrementality entirely and scaling spend on attributed ROAS alone, which can quietly inflate a program that is mostly harvesting organic demand.
A short checklist before you scale spend
- Run a branded-search holdout test before assuming defense spend is incremental.
- Negotiate clean-room access and reporting transparency before signing, not after.
- Separate always-on lower-funnel budgets from upper-funnel test budgets so you can read each clearly.
- Benchmark each network on its own terms rather than against a single blended ROAS target.
- Revisit creative and audience segments quarterly, because auction dynamics and competition shift fast.
The fourth recurring mistake is organizational. Retail media often falls between the trade marketing team that owns the retailer relationship and the digital team that owns media buying, and that split leads to fragmented strategy and double-counted results. Brands that win tend to put one team in charge of the full retail media function and give it a clear mandate.
Tools, partners and vendors worth knowing
A brand rarely operates retail media with native network tools alone. A layer of independent technology and service partners has grown up to manage bidding, normalize reporting across networks, and verify performance. Knowing the categories of vendor helps a team decide what to build in house and what to buy.
The major categories of partner fall into a few buckets. Each solves a different problem, and most mature programs end up using more than one.
| Vendor category | Problem it solves | When you need it |
|---|---|---|
| Bid management platforms | Automate and optimize bids across many networks | Once you run several networks at scale |
| Reporting and analytics tools | Normalize metrics into one comparable view | As soon as you use more than two networks |
| Measurement and incrementality partners | Verify true causal lift independently | Before scaling any significant budget |
| Creative and content services | Produce and adapt assets per network spec | When creative volume outpaces in-house capacity |
| Full-service retail media agencies | Run end-to-end strategy and execution | When you lack in-house expertise to start |
The ad-technology businesses serving this space have themselves become attractive enough to reach public markets, a sign of how much capital is chasing the category. The Liftoff Mobile listing was one recent example of investor appetite for retail and commerce ad-tech, which we noted in our coverage of the ad-tech IPO that raised $437m. For brands, the lesson is that the partner ecosystem is well funded and competitive, so there is no need to accept opaque tools or weak reporting.
Choosing partners comes down to transparency and fit. Favor vendors that integrate with the networks you actually use, that surface incrementality rather than only attributed ROAS, and that let you keep ownership of your data and learnings. The same demand for measurable, verifiable performance is reshaping how brands approach automated and AI-driven buying, a theme we explore in our piece on the executive signals pointing to agentic checkout before the 2026 holidays.
How to build a retail media plan in 90 days
A practical entry into retail media does not require a large team or a huge budget, but it does require sequence. Trying to launch on six networks at once guarantees thin, unreadable results, so the smarter path is to start narrow and expand on evidence.
In the first 30 days, pick the two or three networks where your products actually sell well, set up clean campaign structures that separate branded, category and competitive terms, and establish baseline reporting. Insist on clean-room access from the start, because retrofitting it later is far harder. The goal of this phase is clean data, not scale.
In days 30 to 60, run your first incrementality tests, particularly a branded-search holdout, and begin reallocating budget away from placements that only harvest organic demand. In days 60 to 90, expand the surfaces that proved incremental, layer in off-site audiences for demand creation, and formalize a quarterly review cadence. Treat the whole program as a continuous experiment rather than a fixed media buy, and let measured incrementality, not reported ROAS, decide where the next dollar goes.
For brands operating across both physical and digital retail, the in-store dimension is where the next wave of growth and measurement investment is concentrated, and aligning your retail media plan with that shift keeps the strategy current. The connective tissue across all of this is the wider marketing playbook we lay out in our pillar on retail marketing in the age of AI search and social commerce, which puts retail media in the context of every other channel competing for the same budget.
Frequently asked questions
What is a retail media network in simple terms?
It is the advertising business a retailer builds on top of its own store, website, app and shopper data. The retailer sells sponsored placements and audience targeting to the brands that already sell through it, then reports the sales those ads drove. Amazon Ads is the largest example, but Walmart, Target, Kroger and many others now run their own.
How is retail media different from search or social advertising?
The key difference is data and measurement. Retailers know what shoppers actually bought because purchases happen on their own platforms, which lets them tie an ad directly to a sale through closed-loop attribution. Search and social platforms can target intent and interest, but they rarely see the final purchase the way a retailer does.
Which retail media networks are the biggest in the US?
Amazon Ads is by far the largest, generating well over $50 billion a year. Walmart Connect is the strongest challenger, followed by Target’s Roundel, Kroger Precision Marketing, Instacart Ads and a growing list of grocers, delivery platforms and specialty retailers. The right choice depends on where your products sell and how mature each network’s data and tools are.
What is the difference between ROAS and incrementality?
ROAS is the revenue attributed to a campaign divided by its cost, but it often includes sales that would have happened without the ad. Incrementality measures only the additional sales the campaign actually caused, usually through a holdout test. Incrementality is the metric that should drive budget decisions, because a high ROAS can mask spend that is not generating new demand.
What is a clean room and why does it matter?
A clean room is a privacy-safe environment where a brand and a retailer can match data sets without either side seeing the other’s raw records. It lets a brand verify campaign performance against its own customer data and analyze true incrementality. Demanding clean-room access is one of the clearest signs of a mature, well-governed retail media program.
How much should a brand budget for retail media?
There is no fixed percentage, but most brands begin by shifting a portion of existing trade promotion and digital media budgets rather than adding net new spend. Start with the two or three networks where your products sell best, prove incrementality, and let measured results determine how fast you scale. Beginning small and expanding on evidence beats committing a large budget across many networks at once.
Is in-store retail media worth it?
In-store media is one of the fastest growing parts of the category because it reaches shoppers at the moment of purchase and is increasingly tied back to loyalty data. The main challenge is isolating its effect cleanly, since many factors influence an in-store decision. It is worth testing for brands with strong physical retail distribution, especially as measurement linked to loyalty programs improves.
Do small brands benefit from retail media or is it only for large advertisers?
Small brands can benefit, particularly through self-serve sponsored product campaigns that require modest budgets and produce fast feedback. The advantage for smaller advertisers is precise targeting at the point of purchase without the scale needed for traditional advertising. The constraint is bandwidth, since managing campaigns well takes time or a capable partner.
How do I compare performance across different retail media networks?
Comparison is genuinely hard because each network uses its own attribution rules and metrics. The practical approach is to normalize reporting into a common framework, prioritize incrementality over attributed ROAS, and benchmark each network against its own baseline rather than a single blended target. Reporting and analytics vendors exist specifically to make this normalization manageable.