Retail media was built for shoppers who scroll. Shopping agents read product data instead.

Retail media is one of the best businesses in advertising for a simple reason. The ad sits at the point of purchase. A shopper types "dish soap" into Amazon or Walmart, and the first products they see are the ones a brand paid to put there. The retailer knows who bought what, so it can prove the ad worked.

That model assumes a person is looking at the screen.

On November 4, Reuters published an analysis asking what happens when that person is replaced by software. Its reporters, Juveria Tabassum, Nicholas P. Brown and Arriana McLymore, describe a "conundrum" for Amazon and Walmart: how to let AI agents shop on their sites without undermining the advertising revenue and the customer data that pays for so much else.

We think the question for CPG brands is slightly different and more useful. If an agent does the browsing, what exactly is your retail media budget buying?

The size of what is at stake

The numbers in the Reuters piece explain why retailers are cautious.

Show the data

Year	Value
2025	60.7$B
2026 (forecast)	74.4$B

Source: Emarketer forecasts, as reported by Reuters (Nov 4, 2026). Reuters reports brands are expected to spend 22.6% more on retail media in 2026 than 2025's $60.7 billion. Amazon and Walmart hold a combined 88% share of the 2026 estimate.

Amazon's advertising revenue grew 11% year over year in the third quarter, its slowest pace in nearly four years according to LSEG data cited by Reuters. The segment consistently makes up about 10% of Amazon's net sales. At Walmart, CFO John David Rainey told investors in August that advertising, together with membership and marketplace, accounted for 45% of quarterly operating income growth.

Against that, Morgan Stanley researchers estimated in July that AI agents could account for 10% to 20% of US ecommerce spending by 2030. Nobody knows if that range is right. But even the low end is a meaningful slice of purchases where no human scrolls past a sponsored tile.

Andy Jassy said last week there was no meaningful effect yet on Amazon ads from third-party agents. That is believable. It is also a statement about now.

Two retailers, two bets

The two companies are responding in opposite ways, which is useful for brands because it shows both futures at once.

	Amazon	Walmart
Third-party agents	Blocked Meta's Muse in September; won, then lost on appeal, an order against Perplexity's Comet (case ongoing)	Partnering with Meta's Muse; rolled out agent infrastructure on October 21
How agents shop	Amazon says third-party agents should operate openly and respect a store's decision on whether to take part	Walmart creates a profile for each agent linked to the shopper's account, with approvals and spending limits
Ads in its own assistant	Ads embedded in Alexa for Shopping	Testing ads inside Sparky this year

Source: Reuters via CNA, with Walmart's own announcement.

Notice the last row. Both retailers are putting ads inside their own AI assistants. The bet is that the assistant they control becomes the new search box, with sponsored answers instead of sponsored tiles. Outside agents are harder to sell to, so one company keeps them out and the other lets them in on its terms.

We looked at what that split means for brands in our post on retailers picking sides. The Reuters piece adds the money behind the choices.

What the creative becomes

Here is the part we think matters most for a brand team.

Classic retail media creative is visual and emotional: a hero image, a badge, a headline, a well-placed sponsored slot. Reuters quotes industry practitioners saying that advertising for bots "may rely less on visual flair or emotional resonance and more on specific product data that matches customer queries."

Google's AI Mode is already showing what that looks like. In a Search Engine Land column on November 4, Sarah Stemen describes AI Mode grouping products into categories that Google's AI names and describes, such as "Classic Crewneck Sets," with sponsored and organic products in the same carousels. The category descriptions appear to be generated by Google. The product titles seem to come straight from advertisers' feeds. Her summary: "Your product feed is becoming the creative brief for how Google's AI presents what you sell."

And the paid slots are multiplying. Search Engine Roundtable reported the same day that AI Mode answers have been spotted with up to four ads, where one or two had been typical.

Put those together and a pattern appears. In an AI answer, a sponsored placement still exists, but the words around it are written by the AI, from your data. A feed that says "Protein Bar, Chocolate, 12 Count" gives the model nothing to work with. A feed and page that say "12 g protein, 1 g sugar, no sugar alcohols, gluten free, 12 bars" let it place you under "low sugar options" and explain why.

That same data shapes the unpaid picks too. Stemen makes the point directly: improving feed quality can strengthen visibility across both paid and organic results.

What we think will happen to the budget

We don't think retail media budgets collapse. The Reuters piece has sensible voices on that. Morningstar analyst Brett Husslein calls this "an evolution rather than an existential threat." Mondelez's Andrew Lederman said in April that retailer media sites still drive the vast majority of volume for the company. Amie Owen of Flywheel says her firm is not recommending that clients shift budgets overnight.

But we think three things change in what the money buys.

1. More of the spend moves into the retailer's own assistant

If shoppers increasingly ask Rufus, Alexa for Shopping or Sparky instead of typing a keyword, sponsored answers become the main inventory. That is still retail media, with a different interface. Brands should ask their retail media teams now what reporting exists for assistant placements, and how a sponsored answer is labeled.

2. Pricing and terms change before formats do

Dan Marc of Footprints AI told Reuters he expects multiyear deals for the largest advertisers and more discounting on digital retail media. Katherine Black at Kearney goes further, imagining a "bidding system" in which agents search at a shopper's price point and retailers set ceilings and floors. If anything like that arrives, price and pack data become part of the ad itself.

3. The cheapest lever is the product record

The part of AI visibility a brand controls most directly is the data: titles, attributes, ingredients, claims, sizes, images with readable text, and the same facts on every retailer page. Reuters notes that Procter & Gamble, Nestle and L'Oreal have all said they are working to make product information digestible for AI agents. Fixing that costs time rather than media dollars, and it serves the paid slot, the organic pick and the outside agent at once.

When the shopper is software, the product record is the ad.

How we would split the work

None of this calls for a new team. It calls for moving some attention from the slot to the substance.

• Keep the retail media plan, but ask for assistant-placement reporting at every retailer that offers it. Our AI Mode ads checklist covers the Google side.
• Fund a product data pass across Amazon, Walmart, Target and your own site, checking that titles, variants and claims match. Tip 5 has the basic checklist.
• Decide your agent stance on your own store. Walmart's approach of letting agents in with a profile and limits is one model; tip 33 covers the choice for smaller brands.
• Watch the answers, not only the ads. Run the questions shoppers ask in your category through the main assistants and the retailers' own, monthly, and note who is named and why. Some teams use tools such as SearchPanel from XENA Intelligence for this; a fixed list of questions in a spreadsheet will do the job for a small range.

What would change our mind

If retailers start reporting that agent-driven orders respond to sponsored placement the way human clicks do, the old model survives with a new face. If agent shopping stays a niche, as some current survey data suggests, budgets will barely move. We will watch for both.

For now, we think the safer bet for a CPG brand is the boring one. Keep buying the shelf space that still works. Spend real effort on the product facts that an AI reads before it decides which products to describe. Those facts will be read whether you paid for the slot or not.