WARC Media's Future of Commerce Media 2026 report, out this month, put a number on something the industry has been quietly doing for a year: global retail media spend crossed $200.4 billion in 2026, on its way to $223.4 billion in 2027. In the same window, Ipsos ran simulated shopping trips on Amazon and Walmart and found that identical ads encode into memory 47% worse when shown inside the retailer's own platform than when shown off-site. Read those two findings together and the story stops being a growth story. The channel scared budgets ran to as their 'safe, measurable' harbor is quietly the one shoppers remember the least.

The Budget Didn't Shrink. It Hid.

IAB's 2026 Outlook Study found 90% of buyers still concerned tariffs will hurt ad spend, but the share who actually cut budget in response fell from 45% last year to 30% this year. That gap is the real story: fear stayed high, action dropped by a third. The money didn't disappear. It moved into retail media, now 20 to 35% of a typical CPG ad budget, up from under 5% in 2020, growing 26.1% this year, the fastest of any channel IAB tracks. US retail media alone is on pace for roughly $71 billion in 2026, up about 18% year over year, a bigger single-year jump than search or social posted at the same stage of their growth curves. Retail media became the place nervous marketers could point to a dashboard and say the spend was justified. Nobody got fired for buying a channel with a conversion number attached to it, that number is exactly why it absorbed the fear.

Why a Shopping Brain Skips Your Ad Before It Reads It

The 47-point memory gap isn't a bad-design problem, it's a cognitive one. A shopper on a retailer's site is already executing a task, find the item, compare the price, check the reviews, and task-focused attention filters out anything that isn't load-bearing for that task, the same selective-attention effect that makes you miss a friend's text while parking a car. Psychologists call this a form of inattentional blindness: the visual system still registers the ad, the eye tracks past it, but encoding into durable memory requires spare processing capacity the task has already claimed. Off-platform, in a feed or a CTV break, there is no task competing for that budget, so the ad gets a cleaner shot at memory. Retail media wins the click because it sits at the moment of purchase intent. It loses the brand because purchase intent is the one mental state with the least spare capacity for anything else, including the ad that just won the auction.

What $200 Billion Actually Buys

Run the math on that 47-point gap and the sticker CPM stops meaning what you think it means. If half your impressions are being encoded, your real cost per remembered impression is closer to double the quoted rate, before you even get to concentration. WARC found Amazon holds 78% of US retail media spend, Walmart 7.5%, every other network splitting the remaining 14.5%. Strip Amazon out of the $200 billion headline and the rest of the category is forecast to grow just 9.8% in 2027, the slowest rate WARC has recorded since it started tracking the sector. Meanwhile Amazon, Home Depot, Macy's and Walmart are each running more than 20 ads per page. The channel is getting more expensive, more crowded, and less remembered, in that order, at the same time.

WARC's own framing says it plainly: the headline number looks healthy because one company is carrying it. Everyone else in the category is already slowing down.

What This Means for Your Q4 Budget Meeting

The instinct under tariff pressure is to defend the number you can point to. That instinct is exactly what built a 78%-concentrated, memory-poor category into the fastest-growing line on the media plan. Three moves before you lock Q4:

• Split retail media reporting by network, not by category total, an Amazon dashboard number is hiding what Walmart, Instacart, or Target Roundel are actually doing for recall, not just clicks.
• Track cost per remembered impression, not cost per click, on at least a sample of retail media placements. A 47-point encoding gap changes which channel actually wins a straight ROI comparison.
• Keep one off-platform, brand-memory channel funded through the caution cycle, CTV or social, anything not competing with a shopping task for attention. Retreating entirely into retail media doesn't reduce your risk. It concentrates it, into the same crowded platform everyone else scared this quarter is also running to.