Retail media networks are the channel everyone agreed to be excited about. Spending keeps climbing, and almost every large retailer now runs, or wants to run, its own media network. Ask the marketers who buy it and a quieter story comes out: many retail media networks are walled gardens with limited reach, uneven execution and measurement that holds up only if you do not look too hard. The growth is real. So is the disappointment.

This piece sets out what retail media promised, where the walled-garden model broke, and what is replacing it.

The promise was intent and proof

The pitch was clean. High-intent shoppers, reached at the point of purchase. Closed-loop measurement that ties an ad to a sale. Better signal as third-party cookies disappear. On paper, retail media fixed digital advertising's two oldest problems at once: weak intent and weak measurement.

Where it broke

Most networks ended up as isolated platforms with capped reach, each asking brands to learn another buying system. Execution varied from one retailer to the next. Finance teams started asking questions marketers could not answer with confidence:

  • Is this spend incremental, or expensive retargeting of people who would have bought anyway?
  • Does it scale, or does it cap out as soon as the budget goes up?
  • Why does performance fall apart as spend rises?

The honest answers point at a structural flaw, not an execution problem.

The mistake was building a platform instead of using the asset

Retailers mistook building an ad platform for owning leverage. They assumed the software was the valuable part: the targeting interface, the dashboard. It was not. The asset was always the first-party purchase data, and the demand signal that comes from knowing what people actually buy.

By turning every retailer into a small walled garden, the industry split budgets across many incompatible systems, none with the scale or the optimisation of the tools brands already use. A brand running campaigns on ten retail media networks is not running a media strategy. It is running ten operational headaches, each with its own login, its own reporting logic and its own measurement claims. The leverage retailers thought they were building became drag, for brands and retailers alike.

The model replacing the walled garden

The smarter path inverts the logic. Instead of asking brands to learn another buying system, the retailer routes its first-party signal and inventory to where buyers already work. It keeps what is its own, the data and the demand signal, and stops trying to be a software company.

When this works, the incentives line up:

  • Brands get the high-intent signal and closed-loop measurement without the drag of ten separate systems.
  • Retailers earn from their data without the cost and distraction of running an ad platform they were never built to run.
  • Budgets move faster to where performance compounds, instead of stalling inside fragmented networks.

The value was never the walls. It was the data behind them, and retail media starts to deliver what it promised the moment retailers stop confusing the two.

Retail media is joining the wider stack

That is also why retail media is getting harder to separate from the rest of advertising. As commerce signal flows into larger buying environments, and now into discovery led by agents, the line between a retail media network and the general advertising stack blurs. Retailers who treat their data as a portable asset, not a captive one, are ready for that. Those still building higher walls are optimising for a model that is closing.

The same pressure runs through media buying more broadly. With a shared protocol such as AdCP, any buying agent can talk to any selling agent, so a seller no longer needs a buying interface of its own to be bought. It needs its inventory, prices and rules described in a structure buyers can read.

What this means for brands in Southeast Asia

Retail and commerce platforms across the region are building media offerings at pace, and the temptation to chase each new network as it launches is strong. The discipline worth holding is simple: judge each retail media opportunity by the quality of its first-party signal, and by whether that signal can be measured and acted on. Not by the polish of its dashboard.

A few questions help:

  • Is the purchase signal genuinely first-party, and how fresh is it?
  • Can results be measured against a baseline you trust, not only the network's own report?
  • Does the signal flow into the tools your team already runs, or does it need one more login?

A network that routes real purchase data into the way you already buy is worth more than a standalone system with an impressive interface and reach you cannot verify. Owned signal beats rented complexity.

Where Oderra starts

Oderra applies the same logic to premium media. Fanera, the selling agent from Oderra, answers buyers' briefs at the seller's prices and within its rules, over a shared protocol, so the seller does not have to build a buying system of its own. Skopa, the buying agent from Oderra, sends one brief to every matching seller and brings the answers back in one structure. People on both sides confirm what commits money or inventory, and both read delivery in the same record.