The Maze: Retail media still owns the room. It just no longer owns the category. EMARKETER expects retail networks to fall from 87.0% of US commerce-media ad spending in 2026 to 82.6% in 2030, while nonretail networks rise from 13.0% to 17.4%. That is a 4.4-point shift inside a market forecast to reach $142.07 billion. The money is moving toward companies that see a commercial decision without necessarily owning the checkout.
Dominance and direction can coexist. Retail remains more than four-fifths of commerce-media spending in 2030. Its dollars can keep growing even as its share falls because the whole market is expanding. The forecast therefore does not announce retail media's decline. It describes a larger competitive set. Retailers keep the deepest purchase relationships, but delivery, finance, travel and comparison platforms are turning their own signals into credible media products.
The new inventory is a decision moment. Uber and DoorDash can observe local movement and meal intent. American Express, Chase and PayPal can see transactions across merchants. Airlines and hotels know when travel is being planned. Comparison services such as idealo sit near product choice even when another company completes the sale. These businesses do not need to copy Amazon. They need to prove that their signal identifies a valuable audience or moment that a retailer, search engine or social platform cannot reproduce as precisely.
A signal is not yet a network. EMARKETER's follow-up analysis says different network types need different playbooks. Financial networks need richer commerce context. Travel networks need more frequent engagement. Last-mile platforms must become useful partners rather than another toll booth. Advertisers want open buying connections, comparable metrics, new-to-brand reporting and credible incrementality. Data opens the door; operational usefulness keeps the budget.
More networks create a portfolio problem. Advertisers in a 2025 McKinsey survey used a median of about six commerce-media networks, up from four a year earlier. Two-thirds used at least five. Yet only 3% said networks measured audience incrementality very accurately. The research points to the next bottleneck: buyers must compare overlapping audiences, inconsistent attribution windows and incompatible definitions across a growing stack. Closed-loop reporting is useful, but a sale observed by a network is not automatically a sale caused by its ad.
Why it matters: Commerce media is becoming less about where the transaction ends and more about who owns a distinctive signal before it happens. Retailers retain scale, purchase data and high-intent inventory. Nonretail challengers bring cross-merchant behavior, payment activity, travel plans and local context. The winners will not be every company with first-party data and an ad-sales deck. They will be the few that convert an exclusive signal into reach, simple activation and evidence of incremental sales. For advertisers, that means commerce media needs portfolio governance: one budget logic, one measurement standard and a hard answer to what each network knows that the others do not.
Sources: EMARKETER forecast | EMARKETER network analysis | McKinsey
