U.S. retail media advertising was forecast to reach $60.81 billion in 2025, according to eMarketer's retail media coverage. That's the number that changes how you should think about retail media networks. This isn't a side program run by the trade team, and it isn't just sponsored listings under a search box.

A retailer that owns the shelf, the search results, and the receipt can sell you something a normal publisher can't. It can show an ad and then tell you whether the shopper bought. The catch is that the same retailer also grades its own homework, so you need to understand the buy and the measurement before you trust the score.

Retail Media Networks by the Numbers

Start with a plain definition. A retail media network is a retailer selling attention on its site, app, stores, and offsite channels using its own shopper and transaction data. You're not buying generic reach. You're buying a chance to interrupt a shopping trip that's already underway, plus the retailer's record of what happened next.

That combination of inventory and receipts explains the speed. Retail media was described as the fastest growing tracked U.S. ad channel in 2024 at 20.4%, with projected growth of 20.2% in 2025, according to the eMarketer coverage linked above. The same estimate characterized retail media as adding more new advertising dollars than Meta and Alphabet combined, which tells you this isn't only money shuffled between the old platforms.

What the Table Is Really Saying

U.S. retail media search ad spending was forecast to approach $40 billion in 2025, according to eMarketer's retail media forecast update. In a 2023 survey published in December 2023, 87% of U.S. respondents said they used retail media networks, while 83% planned to increase RMN spending, according to DoubleVerify's report on retail media networks.

Metric	Figure	What it tells a buyer
U.S. retail media ad spend, 2025	$60.81 billion	New dollars, not only a shift off Meta and Alphabet
Growth in 2024	20.4%	Fastest growing tracked U.S. ad channel that year
Projected growth in 2025	20.2%	Expansion stayed elevated after the 2024 surge
Retail media search ad spend, 2025	About $40 billion	Onsite search is still the bulk of the category
Platforms cited by industry estimates	More than 50	Most advertisers will only fund a short list
U.S. respondents already using RMNs	87%	The channel is mainstream, not a pilot
Respondents planning to increase RMN spend	83%	Budgets were still moving in
Marketers with 10% to 39% of budget in retail media	57%	For a majority it is already a real line item

Read those rows as a concentration decision, not as a land grab. Search dominates because intent is clearest there, so sponsored listings will lead almost every sales pitch you hear. Platform count pushes the other way. One industry estimate cited more than 50 retail media platforms, yet most advertisers only want a limited number of partners, because every new network adds contracts, specs, dashboards, and another definition of success.

The Publisher Comparison

A news site can count an impression and maybe a click. A retailer can connect exposure to a basket, a repeat purchase, or a store transaction when identity and loyalty data line up. That's why buyers tolerate closed systems and messy reporting. The receipt feels like proof.

Practical rule: Fund the few networks whose shoppers already buy your category, then judge them on whether they create sales you wouldn't have had anyway.

Don't measure seriousness by how many logos you can activate. A short list gives you enough volume to test creative, negotiate placement level reporting, and learn one retailer's shopper behavior properly. Spreading a thin budget across a dozen networks leaves you with twelve dashboards and no test big enough to believe.

How Retailers Turn Stores into Media

Brands have paid retailers for position for decades. A shelf fee bought a physical slot at eye level or space on an endcap for a promotion. Retail media networks charge for attention on that same shopping trip, only now the slot might be a search result, a product page banner, a streaming ad, or a screen by checkout, and the retailer invoices it like media.

The modern category is generally traced to Amazon's introduction of sponsored placements in its own search results in 2012, according to Stratrix's history of how stores became advertising businesses. That move proved a retailer could monetize high intent moments inside its own store. The model then spread from the owned website into offsite digital channels and physical stores.

A diagram illustrating how retail media networks transform physical shelf space into measurable digital advertising inventory.

The Four Step Loop

Say a shopper searches a grocery app for oat milk, adds a rival brand, and checks out with a loyalty ID. The retailer now holds several useful records from one trip. Here's how those records become inventory you can buy.

Collect first party signals: The retailer logs searches, product views, loyalty activity, and transactions from its site, app, and stores.
Package audiences and placements: It turns those signals into segments and ad slots, such as category buyers, lapsed buyers, or shoppers viewing a competitor page.
Sell the exposure: You bid in an auction or buy a managed placement, depending on the format and network.
Match exposure to sale: The retailer connects the ad event back to a transaction record and reports a result.

Both sides get something specific. Retailers gain media revenue and a larger share of supplier budgets without selling another unit of inventory themselves. You gain access to people who are already on a buying mission, plus a transaction record instead of a click that disappears into a browser session.

Store operators feel the operational side of that loop at checkout, where a clean transaction record matters as much as the ad itself. If you're on the retail side of the counter, a practical checkout guide for shop owners covers the point of sale choices that produce usable sales data in the first place. Screens in aisles and near registers also borrow from an older discipline, which is why digital out of home advertising principles help when you judge placement quality and dwell time.

Where the Loop Cracks

Growth created a definitions problem. Impressions, views, audiences, conversions, and return on ad spend didn't mean the same thing from one network to the next, so a result that looked strong in one dashboard couldn't be compared cleanly with another retailer's report. In 2024, the Interactive Advertising Bureau and Media Rating Council released Retail Media Measurement Guidelines covering data collection, reporting, viewability, and audience measurement across onsite, offsite, online, and in-store advertising, according to the Stratrix history linked in this section.

Buyers felt that gap directly. IAB Europe reported that 70% of buyers viewed the lack of standards as a barrier to retail media investment, according to the same Stratrix account. In Australia, IAB's 2024 survey of 134 advertising decision makers found that 37% of retail media investors had allocated a significant share of their media or marketing budget to the channel over the preceding 18 months, up from 26% in 2023, while 69% identified return on ad spend as a top priority, according to that account.

Treat that Australian result as a caution, not a U.S. benchmark. It shows buyers prioritizing a metric that networks didn't define consistently. The loop only deserves your trust when two records hold up, the impression record that proves a real person had a chance to see the ad, and the transaction record that proves what sold.

The Inventory You Are Actually Buying

A retail media buy isn't one product. It's a stack of placements with different intent, different creative rules, and different measurement risks. Read the stack from the search box outward, because each step moves you farther from a shopper who has already typed what she wants.

A diagram illustrating how retail media networks incorporate search, sponsored products, product pages, and display ads to shoppers.

Sponsored products sit in search results and on product pages. You're buying position at the moment of highest intent, which is powerful and dangerous at the same time. If your product already ranks first organically for its own name, a sponsored click may tax a sale you would've won for free.

Onsite display covers homepages, category pages, and product pages while someone browses. Intent is softer than search, and quality swings hard by placement. A banner above the fold on a category page and a banner buried below reviews shouldn't be averaged into one cheerful click rate.

Beyond the Retailer Site

Offsite display and video rent the retailer's audience on the open web. You're buying the retailer's knowledge of who buys the category, delivered through the same pipes you'd use for other audience buys, so the mechanics in this guide to programmatic ad buying apply to pacing, frequency, and overlap checks. Ask how the audience is matched and how much of it you already reach through Meta, Google, or another retailer.

Connected TV uses that commerce identity in streaming. The screen is bigger and the story can breathe, but the gap between exposure and purchase gets longer. Last touch reporting will undercredit some real effects and overcredit others, so judge connected TV on lift over a sensible window rather than on same day clicks.

Format	Where it runs	Shopper moment	Measurement caution
Sponsored products	Onsite search and product pages	High intent query or browse	Can cannibalize organic rank and branded search
Onsite display	Homepage, category, and product pages	Mid-shop browse	Quality varies sharply by placement and viewability
Offsite display and video	Open web using retailer audiences	Before the site visit	Audience match and overlap with other platforms
Connected TV	Streaming bought with commerce data	Awareness with a retail identity	Long conversion windows and weak last touch reads
In-store screens, audio, checkout	Aisles, endcaps, and checkout	Last steps before purchase	Count opportunity to see and zone, not device plays
Sampling and digital shelf	Physical store zones	Trial at the shelf	Needs zone level sales lift, not impression totals

The In Store Rule Most Buyers Miss

In-store media includes screens, audio, checkout placements, sampling, and digital shelf units. It's the layer most explainers skip, and it's where impression counts get slippery. Proposed IAB Europe standards clarify that an in-store impression should be anchored to footfall or opportunity to see rather than merely counting ad plays, and distinguish measurement by store zone, according to the IAB framework for maturing in-store media measurement.

That distinction matters in a real aisle. A screen can play your ad to an empty aisle at dawn and log a perfect delivery record. What you want to know is how many shoppers passed through that zone, whether they had a realistic chance to see the screen, and whether sales in that zone moved against comparable stores.

More screens and more reported impressions don't automatically mean more effective media.

Use the same comparison concepts across every row of the table. Ask for reach, opportunity to see, viewability, sales lift, and cost per incremental buyer, then add the commercial terms to that same comparison.

Fee structure: Ask whether you pay per click, per impression, or as a share of media, and what platform or data fees sit on top.
Placement guarantees: Ask which positions, zones, or dayparts are guaranteed and which are only targets.
Audience overlap: Ask how the network deduplicates shoppers across onsite, offsite, connected TV, and in-store formats.

Creative fit belongs in this audit too. A TikTok style talking head that works with sound on and captions in a feed often fails in an aisle, where the product must be readable in a second, the offer must work without sound, and the loop has to land before the cart moves on. Map your current spend onto the table rows, and pause any row you can't measure at placement or zone level until the network answers those questions.

Why Reported ROAS Misleads Buyers

A lower ROAS from an independent lift test can be more useful than a higher number on a retailer's dashboard. The dashboard usually estimates correlation, sales that happened after an ad. A well designed test estimates cause, sales that happened because of the ad.

Incrementality is the gap between those two ideas. If a shopper searched your brand name, clicked your sponsored listing, and bought the product she already buys every week, the platform records a clean attributed sale. Your ad may have contributed nothing except a click fee on a trip that was already decided.

A comparison chart showing how independent lift tests provide more accurate ROAS than retailer dashboard attributed sales data.

The IAB and MRC guidelines recommend randomized controlled trials, synthetic controls, or matched market tests to isolate sales caused by advertising from demand that would have occurred anyway, according to the IAB and MRC retail media measurement guidelines explainer. That recommendation should reset your default question. Don't ask which campaign has the best ROAS tile. Ask which design would convince a skeptic that the ad changed behavior, which is the same discipline behind sound attribution modeling choices.

Pick the Design That Fits the Constraint

Method	Use it when	What it estimates
Randomized controlled trial	Shoppers or markets can be randomly split	Difference in conversion or revenue versus control
Matched market test	User level randomization is unavailable	Lift versus similar untreated markets
Synthetic control	Only a few markets are treated	A weighted counterfactual from untreated markets
Media mix modeling	Spend is always on across channels	Contribution after other media and seasonality
Audience holdout with creative splits	The question is which hook works	Incremental lift by hook, angle, or segment

A randomized controlled trial gives the cleanest read because exposed and control groups start out alike, so the difference in revenue estimates lift. Matched markets and synthetic controls are practical fallbacks when a retailer can't randomize individual shoppers, which is common for in-store screens tied to a whole location. Media mix modeling suits always on budgets spread across channels, where turning everything off for a clean test isn't realistic.

Build the Test in This Order

Predefine the setup: Lock the eligible population, treatment window, conversion window, primary outcome, exclusion rules, and minimum detectable effect before launch, as the IAB and MRC explainer linked in this section recommends.
Retain the raw records: Keep exposure, eligibility, and transaction data so someone outside the sales team can audit the result later.
Filter junk first: Remove invalid traffic, including bot generated impressions and clicks, and tie attribution eligibility to viewable impressions.
Keep placement detail: Preserve placement ID, format, device, timestamp, and exposure status instead of accepting one network level total.
Split creative inside treatment only: Randomize hooks within the exposed group so each variant is judged on lift, not on the cheapest attributed conversion.

Those steps protect you from a specific failure mode. Last touch and view through attribution overcredit ads shown to shoppers who were already likely to buy, and network level totals hide whether results came from strong placements or from duplicated, unviewable, low quality delivery. Placement level, auditable data lets you compare a hook in sponsored search against the same idea in offsite video without mixing measurement artifacts into the creative verdict.

Report the reads dashboards blur together. Separate new to brand buyers from repeat buyers, because stealing a repeat purchase from your own organic listing isn't growth. Track halo sales on related SKUs in the basket, and watch for cannibalization of organic search or retailer driven traffic that makes total brand sales flat while attributed sales climb.

Ask which lookback window produced the number in front of you, since a longer window sweeps in more purchases that may have little to do with the ad. Then send your partner one direct question before you scale. Which incrementality design will you accept, and can we see placement level viewable impressions before the next budget increase?

Campaign Patterns from Leading Networks

You don't need a ranking of every network with a sales deck. You need patterns you can copy, because each major network teaches a different lesson about intent, identity, and where measurement breaks. Steal the decision each pattern forces, not the logo.

Amazon is the original closed loop. Sponsored search works best when the product page already converts, with reviews, price, imagery, and stock doing the selling after the click. The decision it forces is uncomfortable. Before you raise bids, check whether you're paying to intercept shoppers who searched your brand and would've bought anyway, and fix a weak product page before buying more traffic to it.

Walmart Connect is the omnichannel test. A large online grocery and general merchandise audience sits next to a national store fleet, so onsite, offsite, and in-store activity can all touch the same household. The useful question isn't whether each format reports sales. It's whether search ads plus aisle screens grow total sales of the SKU, or move a purchase from the store trip to a pickup order while you pay twice for the privilege.

Say a laundry brand runs search for a new scent while screens promote it near the detergent aisle. Store level sales in test locations versus matched control stores tell you more than two dashboards claiming the same basket. If total category sales for the brand don't move, you've bought channel shift, not growth.

Target Roundel is the offsite lesson. Guest and loyalty data pushed into display, video, and social placements is built to create demand before a shopping trip, not to harvest a search that already exists. Judge it on new to brand buyers and later store or site sales among exposed households. If you grade an offsite demand campaign with same day last click rules, you'll kill the very activity that feeds future search volume.

Independent sellers learn a smaller version of that demand lesson outside retail media entirely. If you also sell through marketplaces, you can browse Etsy bestseller keyword ideas to see how shoppers phrase problems before a brand name enters the search. That language often makes better offsite hooks than your own product vocabulary.

A grocery delivery network such as Instacart is mission media. Shoppers arrive with a list, a recipe, or a replenishment habit, and sponsored products slot into a basket that's already being filled. Intent is superb and cannibalization risk is just as high. The first question is whether your ad put a new item in the basket, swapped a competitor out, or charged you for the item the shopper reorders every week from her buy again list.

What Shifts Through 2026

Measurement standards from IAB and MRC are becoming the price of entry in serious retailer conversations, so vague dashboard definitions will get harder to defend in a renewal meeting. In-store measurement is still phased, which means zone level opportunity to see will separate credible offers from pitches built on screen counts. Offsite and connected TV keep stretching retail audiences away from the product page, so identity matching and overlap control matter more each quarter.

Fragmentation pushes in the other direction. Brands will work with fewer networks more intensively because each additional partner dilutes test volume and creative learning. The advantage moves to teams that can reversion one proven idea for a search thumbnail, a silent aisle screen, and an offsite video without waiting weeks on a studio, then prove each version inside a holdout.

Choose Networks and Test Creative

Fund a network only when three filters all pass. Its shoppers already buy your category in meaningful volume. It supports an incrementality design you'll accept. It reports viewable results at placement level. If any filter fails, keep your money and fix the gap first, because a bigger budget poured into weak measurement only produces more confident looking noise.

This test comes before expansion, even though retail media networks now take a serious share of budgets for many teams. In a 2023 survey published in 2024, 57% of marketers said retail media represented 10% to 39% of their marketing budgets, according to the Association of National Advertisers retail media announcement. Among 174 marketing professionals surveyed for a 2024 study, 8% identified scalability as a retail media challenge, down from 29% in 2023, according to Modern Retail's coverage of that research.

That shift changes your bottleneck. Adding another network login is easy now. Producing enough good creative variants to learn anything is the scarce input, especially when small retailers and local sellers also need hands on help. If you're building the selling operation behind the ads, you can find support for selling locally while you keep the media test narrow and measurable.

The planning view below shows the choice you're making, networks on paper against the evidence each one can produce.

A professional analyzing retail media network options on paper and a laptop screen for advertising strategy.

Run This Test Before Any Budget Increase

Pick one hero SKU and one primary network: Choose the product with healthy margin, reliable stock, and a product page that already converts organic traffic.
Write five hooks: Cover the problem, the proof, the offer, a comparison against the usual choice, and the main objection shoppers raise in reviews.
Produce many variants, not one film: Build UGC style cuts and photos for each hook, short enough and clear enough to work without sound onsite, in social placements, and on an aisle screen.
Cap frequency on small audiences: Retire winners before fatigue sets in, because a tight category segment burns through creative faster than a broad prospecting pool.
Randomize inside a holdout: Split hooks within the treatment group and judge them on incremental buyers and new to brand sales, using the discipline in this guide to testing ad creatives.
Promote only causal winners: Scale variants that lift sales against the holdout, and kill variants that win only on cheap attributed conversions.

Production speed decides whether that slate exists at all. AI UGC videos generate 4x higher click through rates and 50% lower cost per acquisition than traditional ads, and that kind of volume is what lets you test five hooks instead of debating one script for a month. Brands have used that approach at scale, since MVMT achieved a 47% ROAS improvement with AdCrafty and HelloFresh boosted its conversions by 62% with AdCrafty.

Watch the short walkthrough below before you brief variants, so your team agrees on what a testable hook looks like in practice.

AdCrafty is the production shortcut for step three. We turn a product URL into UGC style video and photo variants in minutes, with many hooks and avatars to choose from, short form exports you can run across placements, and commercial rights that let you use the files as ads with retail and agency partners. You keep the strategy and the test design. We remove the shoot schedule that usually shrinks a five hook plan into a single polished asset.

The Stopping Rule

If the holdout shows no incremental buyers after a fair window, change the hook or the network before you add spend.

That rule protects you from the most expensive mistake in retail media, scaling a campaign because the retailer's tile looks green. A fair window means enough exposed and control shoppers to detect the effect you predefined, with stock and pricing held steady so you don't confuse a promotion with creative impact. If the test does show incremental buyers, move budget out of attributed only placements and into the exact cell, placement plus hook plus audience, that caused the sale.

Document what you learn in plain language your finance partner can audit. State the design, the window, the viewability filter, the new to brand split, and the cost per incremental buyer next to the reported ROAS. When those two numbers disagree, trust the causal one and make the dashboard earn its way back into the budget conversation.

Credible results travel better inside a company than inflated ones. A modest, proven lift on one hero SKU at one retailer gives you a template you can defend at the next network, with the same hooks reversioned for new specs. That's how a short list stays short while your learning compounds.

Build your next retail media test around evidence, not dashboard applause, and give every hook a fair shot inside a holdout. With AdCrafty, you can turn a product URL into a full slate of UGC style video and photo variants in minutes, export them for retailer, social, and offsite placements, and scale only the versions that prove incremental lift. Start your first five hook test today and let causal results decide where your next retail media dollar goes.