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.

Functional Ad Units (FAU) — Ads That Do a Job, Everywhere | Particular Audience

Functional Ad Units compare, advise, reorder and beat a price inside AI chat, publisher pages, social feeds, retailer search and shopping agents — keeping monetisation and margin with the retailer.

Functional Ad Units

Functional Ad Units (FAUs) are sponsored units that do a job — compare, advise, reorder or beat a price — inside the surface the shopper is using.

Not just chat. The next wave of all advertising. FAUs started as ads that work inside AI answers. The same unit runs in publisher pages, social feeds, retailer search and shopping agents. Headless, native commerce, where the ad itself is the shop, is finally arriving.

Feb 2026 — Ads below the answer: OpenAI began testing ads in ChatGPT for US Free and Go users: a headline, short description, image and link, labelled sponsored and shown below the answer. Sources: OpenAI: Testing ads in ChatGPT; OpenAI Help Center: Ads in ChatGPT.

May 2026 — Shopping cards and carousels: A larger image, a call-to-action button and an ecommerce format with price and reviews that can stack side by side. Multi-image ads followed in October. Sources: Digiday, 21 May 2026.

Sep 2026 — Sponsored agents: click to chat: Clicking an ad opens a labelled conversation with a business-sponsored agent, which answers questions and then links to the business's website. In November, free business agents arrived, with sponsored agents as the paid route. Sources: OpenAI, 16 Sep 2026; Digiday, 14 Sep 2026; Digiday, 10 Nov 2026.

Next — Functional Ad Units: Truly agentic ads. Instead of talking and then sending the shopper away, an FAU does the job in place: compares, advises, reorders, beats a price or adds to basket, with the retailer's ranking, live price, stock and supplier funding behind it, on any surface. (Our view of the next stage.)

Compare — powered by Recommendations, Visual Search & See Similar, Merchandiser Control, Search: PA's recommendation engine finds the closest-match items to the one being considered, using visual and text similarity and search, then applies the retailer's upsell or other merchandising rules to hit its business goals.

Advise — powered by Adaptive Transformer Search, Feed enrichment & AEO/GEO: Feed enrichment reasons over each product's attributes and use cases, and Adaptive Transformer Search matches the shopper's need to the right products and advice.

Reorder — powered by Buy Again, Sponsored Products: Buy Again predicts when a shopper is due to replenish from their purchase history and offers the reorder in one tap.

Beat a price — powered by Price Beat, Price Affirmation, Trade Media: Price Beat beats an approved competitor price within the retailer's rules, Price Affirmation confirms when the retailer is already best, and Trade Media lets suppliers co-fund or fully fund the difference.

Surfaces: AI chat, Publisher article, Social feed, Retailer search, Shopping agent.

Assistants only need your feed: AI assistants such as ChatGPT and Claude, and agentic checkout protocols such as UCP and ACP, can make a retailer discoverable from a product feed alone.

Aggregators become ad platforms: Whoever owns the interface can sell the ads. Comparison shopping did it first; AI ad models continue it, building ad revenue lines that used to sit with the retailer.

Merchandising goes missing: A retailer promotes items for profit, cash flow or stock. Agentic discovery serves the shopper only, so that merchandising intent is lost.

Retail media margin leaks: Supplier money follows the interface. Without a standard the retailer controls, onsite retail media and trade funding drift to the aggregator.

Enrich the feed: PA feed enrichment adds attributes, use cases and plain-language descriptions with Adaptive Transformer Search, so agents and LLMs can find and compare your products today.

Carry the offer: Structured prices, Price Beat offers and Trade Media deals travel with the product, so the retailer's best price is visible to agents.

Serve the FAU: Where the surface allows it, the retailer's own FAU answers the shopper and carries its sponsored and supplier-funded value with it.

Measure it yourself: Impressions, funded offers and sales are counted by the retailer, not the interface.

Retailers need to push for open functional ad unit standards. FAUs put monetisation and margin back into how a retailer talks to its customers, wherever that conversation happens.

Own the unit: Insist that ads about your products in AI chats, publisher pages and social feeds are served as your FAU, with your ranking, live price and stock — otherwise the interface keeps the margin.

Carry your offer: Your enriched feed, Price Beat offers and supplier-funded Trade Media deals travel with the product, so agents see your best price.

Measure it yourself: Sponsored units are clearly labelled, and the numbers come from your own systems — the unit's API calls — not another company's analytics tag.

Measurement is API usage, not a third-party tag.

A functional ad unit is served by the retailer, so it runs against the retailer's own systems. What happened is recorded in the calls the unit makes — not in someone else's analytics feed.

The unit calls your APIs: Live price, stock, ranking and basket adds are requests into your systems. Each one is an event you can log against the campaign.

Funded offers settle from your records: Supplier-funded deals reconcile against your own order data, so funding is measured the way media is.

No tag in the middle: The interface doesn't have to report what happened for you to know. You already saw the calls.

PA notifies AdOps and advertisers on pacing; it never changes floors or caps.

Money-flow shares and demo products are illustrative.