Particular Audience Particular Audience
Talk to us

AI for search, personalisation and retail media

A different store for every customer.

Bring what each shopper wants together with what your business needs.

Search, recommendations, retail media and promotions across your digital store and AI shopping experiences.

Is the AI agent about to steal your retail media? Aggregators as ad platforms | Particular Audience

Why AI interfaces that only need a product feed can disintermediate retail media, why onsite retail media may become a mediation layer, and why best price, Trade Media and functional ad units matter.

Our first thesis

Our initial thesis for preserving retail media in agentic commerce was MCP over merchandised search. A retailer's MCP server answers agents from its own search, so retail media ranking and other merchandising controls survive in how the retailer responds to agentic demand.

What we mean by “aggregator”

We borrow the term from Ben Thompson's Aggregation Theory, first published on Stratechery on 21 July 2015. He splits every consumer market into three parts: suppliers, distributors and consumers. Before the internet, distributors won by owning supply. The internet made distribution of digital goods free and transaction costs close to zero, so the winners now own the customer relationship instead. In his words, aggregators “aggregate modularized suppliers — which they often don't pay for — to consumers/users with whom they have an exclusive relationship at scale.” In Thompson's framing the “suppliers” are whoever provides the goods or content. In this article that means retailers and their product feeds, not the brand suppliers who sell to retailers and fund their retail media.

The engine is a flywheel: the best experience “earns them the most consumers/users, which attracts the most suppliers, which enhances the user experience in a virtuous cycle,” with what Thompson calls “strong winner-take-all effects.”

Our view, applied to retail: an AI shopping assistant or delivery app that only needs a product feed treats retailers as interchangeable product sources. It keeps the shopper relationship and sells the ads itself. That is what we mean by aggregators as ad platforms. The rest of this article asks how retailers keep their media and their offer under that pressure.

Source: Ben Thompson, “Aggregation Theory”, Stratechery, 21 July 2015 (https://stratechery.com/2015/aggregation-theory/), checked 15 Nov 2026

Source: Stratechery, “Aggregation Theory” hub page (https://stratechery.com/aggregation-theory/), checked 15 Nov 2026

The counter-argument

It is in the interest of the aggregator as an ad platform to disintermediate a retailer's revenue line.

To do that, it would rather index product feeds and run its own ad business. Comparison shopping already did this, and AI ad models continue it.

Quick-commerce delivery apps are the other, more aggressive, route: they too have become aggregators as ad platforms, taking retail media revenue away from the retailers they list.

A twist worth watching

Perhaps prophetically, one large delivery aggregator ran out of its own ad supply and turned to feeding ad demand onto its retail partners' own properties, with a useful incremental ad margin for the retailer. For retailers protecting their trade dollars this was attractive, much like plugging in a retargeting network to bring incremental ad budgets.

If that holds, the future of onsite retail media is a mediation layer: a gateway that lets many outside ad buyers (brands, DSPs and ad networks) bid for space on the retailer's own site and app, under the retailer's rules. Publisher advertising went the same way, with demand-side and supply-side platforms becoming separate parts of programmatic infrastructure. Retail media may follow.

Most likely both paths run at once, as with Google Shopping today. The open question is who sets the offer the agent sees: supplier funding through Trade Media and Price Beat can set one offer price, fed to aggregators and the retailer's own site alike, because in both paths the shopper or agent compares offer prices, not list prices.

Humans will still shop

This view matters for as long as people shop and visit websites, which I expect they will.

The best bull case I have heard for full agentic adoption is credit cards: first “no way I'll put my card online”, then “I'll give one big store my card”, then “I don't care about typing my card details anywhere”. Given that path and the speed of AI adoption, there is a reckoning coming for ecommerce onsite technology in general. That is why we have positioned strongly into MCP, even though market pull has not really started.

Best price wins the agent

The more I think about what matters in agentic commerce, the more I think it is best price. For any commodity product the agent's job is to find the cheapest option.

That is where Price Beat has the greatest application: a retailer can keep a list price and also publish an offer price that beats approved competitors' prices, winning the shopper's agent, which has no loyalty.

Trade Media: the bigger pool

Pricing and promotions is a supplier funding channel that, by our estimate, is around five times larger than retail media as an addressable market. That is why we think supplier-funded experiences are the future, and why we call it Trade Media.

Ads in AI are already evolving

ChatGPT ads began in February 2026 as a sponsored unit below the answer. By May they had shopping cards that could stack into carousels. In September came sponsored agents: click an ad and chat with the business, then follow a link to its site.

The next stage, in our view, is the functional ad unit: an ad that does the job in place, with the retailer's ranking, price and supplier funding behind it.

Why functional ad units

Agentic discovery serves the shopper, not the retailer's profit or cash-flow merchandising. Retailers need a standard they control that puts monetisation and margin back into every customer interaction. That is the case for functional ad units.

The doomer question: is the retailer just a warehouse now?

If aggregators own the shopper and the agent, is this the end of retailers aggregating brands? Does the retailer become a logistics and fulfilment provider?

The honest answer: yes, if all it does is publish a feed. A retailer that hands over raw product data, and lets someone else rank the products, sell the ads and set the winning offer, becomes an interchangeable fulfilment node. That is exactly what Aggregation Theory predicts for modular suppliers.

It doesn't have to be. Retailers hold four kinds of leverage an aggregator can't easily copy.

Atoms are not bits. Stores, local stock, cold chain, click-and-collect and returns are costly physical friction. Aggregators want software margins and avoid owning warehouses, vans and inventory.

The offer stays with the retailer. Curation moves to the prompt, but an agent is loyal only to the best net offer, and the retailer holds the supplier funding to make that offer. That is the list price versus offer price argument.

Supply the agent can't get elsewhere. Own-brand and exclusive ranges are not modular: an agent can only buy them from the retailer that carries them.

Answer the agent yourself. A retail MCP server and functional ad units keep the retailer's ranking, margin rules and supplier sponsorships inside every answer, wherever the shopper or agent asks.

So the role shifts from catalogue aggregator to commercial orchestrator: owning the ranking, the media, the co-funded offer and fulfilment as an advantage.

This is not the “open internet” argument some ad-tech vendors made, casting themselves as defenders of the open web against walled gardens. Our claim is narrower. It isn't about keeping advertising open. It is that retailers hold commercial assets an agent has to transact through: physical fulfilment, the funded offer and their own ranking. Retailers are not a cause to defend; they are a commercial position to use.

That role is what Particular Audience is built for. PA adds the commerce layer alongside a retailer's existing technology, and works with any website, app or ad server.

How retailers survive and thrive in agentic commerce

Own the atoms: Stores, local stock, cold chain, click-and-collect and returns are physical work that aggregators avoid.

Own the offer: Supplier funding sets the net offer price, and an agent's only loyalty is to the best offer.

Own supply no one else has: An agent can only buy own-brand and exclusive ranges from the retailer that stocks them.

Answer the agent yourself: A retail MCP server and functional ad units keep the retailer's ranking, margin rules and sponsorships inside every answer.

A retailer that only publishes a feed becomes a fulfilment node. One that holds these four becomes the commercial orchestrator.

Logistics provider: Publishes a feed for others; Others rank its products; Others sell the ads; Others set the winning price

Commercial orchestrator: Answers agents from its own search; Keeps its own ranking and merchandising; Sells its own media, including to outside demand; Sets a supplier-funded offer price; Uses fulfilment as an advantage

Retail MCP product search: Answer the agent from your own search

Functional Ad Units: Ads that do the job, with your ranking inside

Trade Media: Turn supplier funding into the offer

Price Beat: Beat, affirm or hold the offer price

External Demand Governance: Sell your own space to outside demand, on your rules

Sponsored Products: Monetise search, browse and product pages

Buy Again: Keep repeat shopping on your own properties