Ask five people what retail media is and you'll get five different answers. The retailer will talk about their ad inventory. The agency will talk about the budget they manage. The consultant will talk about "the third major digital channel after search and social." All of them are true. None of them is enough for a brand selling in LATAM in 2026.

This article is the definition from the other side of the table: the side of the CPG brand that has to execute every day on Mercado Libre, Éxito, Jumbo, Falabella, or Rappi. From that seat, retail media isn't just "another paid channel." It's the visible tip of something much bigger — a four-layer iceberg — and understanding the whole iceberg is the difference between your investment paying off or evaporating.

Let's start with the floor: the classic definition

In its simplest form, retail media is paid advertising bought inside a retailer's digital surfaces. Banners on the ecommerce home page, sponsored search in search results, sponsored products on a category page, off-site display activated with the retailer's purchase data. The brand brings the creative; the retailer brings the inventory and the audience.

The name sounds new but the idea isn't. The paid end-cap, flyers with featured products, special displays at Éxito or Carulla — that was already retail media. Digital simply added three things: first-party data on who buys what, near-real-time measurement, and the ability to activate audiences off-retailer (off-site programmatic). As Jeffrey Bustos puts it nicely in his grandma-friendly guide, it's basically an old recipe with new ingredients.

That definition is fine. It's the floor. The problem starts when you treat it like the ceiling.

Why that definition is incomplete for your brand

Each player defines retail media according to their incentive. The retailer defines it as the inventory they sell. More inventory sold, more high-margin revenue. The agency defines it as the budget they manage. Bigger budget, bigger fee. The adtech platform defines it as impressions served. More impressions, more platform usage.

But the brand isn't buying inventory, or budget, or impressions. The brand is buying one thing: that the shopper puts your product in the cart and completes checkout. And for that to happen, the paid ad is one player — but not the only one, and not even the most important one. The sponsored slot sends the customer to a PDP (Product Detail Page). If that PDP is out of stock, mispriced, missing attributes, low-rated, or invisible organically once the ad ends, the ad spend doesn't pay off.

That's why the brand needs a wider definition. Retail media, from its seat, is the entire system that determines whether the sale happens inside the retailer's ecommerce. And inside that system, paid ads are the visible layer — but there are three layers below the waterline that weigh more.

The iceberg: the visible tip and the submerged base

The analogy is old but it works: an iceberg shows roughly 10% of its volume above water. The other 90% is below, holding up the tip. What you see isn't what matters — it's what holds up what you see.

Retail media behaves the same way. What everyone watches — the retailer dashboard, the campaign ROAS, the CPCs, the slots won — is the visible tip. What almost nobody monitors systematically is the submerged base: real product availability per SKU-store, price relative to competitors at the exact moment of the click, organic position that decides what happens when the ad isn't running, the PDP content that decides whether the human shopper (or the AI agent) can convert that visit.

Brands that only watch the tip pay premium CPCs to send traffic to listings that are out of stock, mispriced, or invisible. Brands that watch the whole iceberg understand that paid ads only pay off when the base is solid. And when the base fails, the ads don't save it — they amplify the damage, because now you're sending frustrated shoppers to a bad experience, at scale.

This is the point no retailer dashboard will ever report: your retail media didn't perform this quarter. It wasn't the bid. It wasn't the creative. It was that your PDP was invisible organically, out of stock in 30% of stores, or priced 12% above the competitor, and paid clicks were landing on a listing that didn't convert. The visible tip took the blame for a problem that lived underwater.

The 4 layers, in reverse order (from base to surface)

Layer 1 — Availability and price (the base of the iceberg). If your product is out of stock or expensive, the other three layers don't matter. The shopper can't buy what isn't there, and they won't pay a premium for what their competitor has 10% cheaper two slots below. This layer is monitored with daily PDP scraping at every retailer-store, measuring stock continuity, relative pricing, and promotion propagation. It's the layer with the most immediate ROI: every hour out of stock is a sale that goes to the competitor. Go deeper: Stockouts, prices, and smart bidding: the real-time retail media play.

Layer 2 — Organic position (the health thermometer). When the shopper searches "almond milk" or "anti-dandruff shampoo," the retailer's algorithm shows them a list. Your brand's organic position in that list is the result of everything you did before: you sold, you got rated, you completed attributes, you held stock, you stayed relevant. It's your vital sign. When it climbs, something's working; when it drops, something broke — and almost always it broke at layer 1 before showing up here. Go deeper: Your organic position in ecommerce is a thermometer: learn to read it.

Layer 3 — PDP content (the layer the AI agent reads). Complete attributes, intent-legible descriptions, images with alt text, structured specifications. This layer decides two different things. One, whether your product activates the retailer's filter facets ("organic," "gluten-free," "low sodium") when the shopper filters — relevant on retailers that still have visible facets (Olímpica, Jumbo Colombia, Mercado Libre). Worth noting that Éxito and Carulla removed filter facets recently, so on those retailers attributes still feed the ranker but are no longer a shopper-facing filtering lever. Two — and increasingly important — whether an AI agent or Google's AI Mode can understand, compare, and recommend your product when the shopper asks a natural-language question. Go deeper: PDP in an agentic world: how to optimize for humans and AI, and specifically on the Google engine: The Google engine and Intent Coverage in the new PDP.

Layer 4 — Paid retail media (the visible tip). Sponsored search, sponsored products, on-site display, off-site programmatic, brand stores, and the agentic ad formats that are emerging. This is the layer everyone measures and the one the retailer bills for. It works when the three layers below are solid. It fails spectacularly when they aren't.

Order matters: nobody should pay for sponsored placements to push a SKU that's out of stock in 40% of stores, or to drive traffic to a PDP with two of twelve attributes filled. And yet that decision gets made every week — because the retailer dashboard doesn't show the layers below.

What dunnhumby says (and what nobody else does)

In April 2026, dunnhumby published Retail Media: Technology, Trust and Retail Experiences Reshaping Shopper Decisions, a survey of 3,000 grocery shoppers in the United States and the United Kingdom. The data confirms what many brands intuited but couldn't defend in committee: the shopper does want retail media, but only when it feels useful.

The key numbers from that study: 91% of shoppers are comfortable with their transactional data being captured, and 9 out of 10 are open to more retail media experiences — provided they're personalized. Shoppers are nearly three times more likely to trust personalized advertising (70%) than generic advertising (25%). And, according to a parallel Tesco Media study cited in the same research, 71% say they enjoy discovering new brands. Michael Schuh, dunnhumby's head of retail media, summarized it this way: "retail media is transforming at an extraordinary pace, but the shopper perspective is often missing".

The subtext matters more than the numbers. When dunnhumby says "personalization with trust," it's saying: retail media only works when the personalized ad leads to an experience that delivers what was promised. If your ad offers a brand and the SKU is out of stock, you broke the trust. If your ad personalizes to someone with a gluten-free diet and your PDP doesn't have the "gluten-free" attribute structured, the next filter loses it. Personalization requires layers 1–3 to be solid. Without that base, layer 4 doesn't build trust — it destroys it.

What this looks like in LATAM

The Latin American landscape filled in fast. According to the Retail Media Report Card LATAM H2 2025 by Mars United, retail media surpassed 10% of the region's total digital ad spend for the first time in 2025, is projected to double to USD $5.2 billion by 2029, and nearly half of the active networks today were born in 2024 alone. Argentina, Brazil, and Mexico account for roughly 85% of regional ecommerce sales.

Today's menu includes, among others: Mercado Libre Ads (the largest regional network at scale, with search, display, and product ads on the marketplace), Cencosud Media (Jumbo, Easy, Paris, on a VTEX stack across multiple countries), Éxito Media (Grupo Éxito in Colombia, also VTEX), Falabella Media (Falabella, Sodimac, Tottus in Chile, Peru, and Colombia), Rappi Ads (delivery in 9 countries, heavily category-based formats), and Walmart Connect (Mexico and Central America). Each with its own auction logic, surfaces, and dashboard. The good news: the supply exists. The bad news: each platform is its own universe, as the industry conversation likes to remind us, and comparing performance across them is still hard. The specific ranker weighting per retailer we cover in Structured attributes and the retailer engine; here we just leave the map.

Independent measurement: the thread crossing all 4 layers

There's a problem that becomes obvious once you see the whole iceberg: every measurement you have comes from the retailer or the agency. The retailer reports the ROAS of its own inventory. The agency reports the performance of the budget they manage themselves. It's the equivalent of the match referee also being the owner of one of the two teams.

It's not that they lie, necessarily — it's that they're measuring only the layer it suits them to report on. Neither will tell you "your organic position dropped three places this week," "your price drifted out of competitor range for 11 days," or "your PDP lost the 'sugar-free' attribute and dropped out of the facet." Those are the things that live underwater and are what determine whether layer-4 retail media is going to perform or not.

That's why a serious brand needs an independent measurement layer — daily scraping of all four layers, at every retailer, without asking the retailer's permission. That's the proposition of ePerfectStore.com: be the external thermometer no retailer or agency controls, and tell you whether the iceberg's base is solid before the tip starts to fail. The full measurement framework is in How to know if your retail media really works: the honest guide for CPGs in 2026.

Sources

Closing: base first, tip later

The playbook this blog series details, in one sentence: build the base — availability, price, organic position, content — and retail media will start delivering what the retailer's dashboard claims it's delivering. Without that base, layer-4 spending finances your own failure.

To go deeper into any of the layers, the two related clusters:

Retail media measurement cluster:

PDP in an agentic world cluster:

What does the base of the iceberg look like for your brand? ePerfectStore.com measures all 4 layers across every Colombian and LATAM retailer, without depending on the retailer's dashboard.

Back to Blog