When Bots Do the Buying, Who Gets the Advertising Dollar?

 

The grocery store, C-store and restaurant may still own the customer relationship—but increasingly, the AI agent may own the decision. That changes the economics of food advertising according to Steven Johnson Grocerant Guru® at Tacoma, WA based Foodservice Solutions®.

For decades, food marketers have understood a simple proposition: get in front of the consumer at the right moment, create desire, influence the choice and convert that choice into a purchase.

Retailers learned to monetize that proposition through circulars, endcaps, displays, loyalty programs, email, apps, sponsored search and, most recently, retail media networks.

Now comes the uncomfortable question:

What happens to the advertising business when the shopper stops doing the shopping?

That is the question being raised by Instacart's announcement that Meta's new Muse AI agent will connect to Instacart's grocery infrastructure. Instacart already has connectors involving ChatGPT, Claude, Gemini and Google's AI Mode. Its own Clementine AI assistant can turn a prompt such as “high-protein dinners for two” into a shoppable cart based on real-time store inventory.

This isn't simply another digital-shopping feature.

It is a potential change in who makes the purchase decision.

And that should make every grocery CEO, C-store CEO, restaurant CEO and CPG marketing executive ask a very different question:

If the bot is making the buying decision, who is the retailer selling advertising to?

 

The grocery industry has spent billions building the advertising machine

The retail media opportunity is enormous.

Instacart says its advertising ecosystem now reaches 55+ million consumers, more than 100,000 storefronts, 2,200+ retail banners and 310+ retail media networks and marketplaces.

GeekWire has been documenting the evolution of this business for years.

In 2022, GeekWire reported on Seattle-based grocery technology company Swiftly, which helped brick-and-mortar grocers collect customer data and monetize it through advertising. At that time, Swiftly said more than 80% of retail transactions still occurred in physical stores and roughly 90% of grocery shopping took place in brick-and-mortar stores rather than through an app or browser.

Then the advertising opportunity expanded beyond the store.

GeekWire reported in 2024 that Seattle-area adtech company Symbiosys was targeting a retail-media market estimated at $45 billion, helping retailers sell advertising beyond their own websites and apps.

And in 2025, GeekWire reported that DoorDash's advertising business had crossed a $1 billion annualized revenue rate, while its acquisition of Symbiosys was intended to expand off-platform advertising capabilities for brands and restaurants.

So the food industry has been building something extremely valuable:

First-party data + consumer intent + media inventory + transaction data + closed-loop measurement.

But AI agents introduce a new intermediary between the brand and the buyer.

 

The advertising problem: The bot doesn't need to be persuaded

This is where the food industry needs to rethink the word advertising.

A human shopper might see:

TYSON CHICKEN — $5.99

and think:

"That's a good deal. I'll buy Tyson."

An AI agent may instead process:

·       price

·       package size

·       nutrition

·       dietary requirements

·       availability

·       promotions

·       previous purchases

·       delivery time

·       retailer

·       product ratings

·       household preferences

·       substitution rules

and then select the product.

That is a fundamentally different purchasing environment.

Deloitte describes the emerging “algorithmic shelf” as an environment in which AI agents increasingly analyze data and verifiable claims rather than responding to traditional marketing persuasion. Deloitte reported in June 2026 that 55% of consumers were starting shopping journeys through large language models, according to its cited research.

Circana reported another important 2026 data point: 25% of consumers surveyed had used an AI recommendation to purchase a CPG product.

And Deloitte's 2026 retail outlook says AI-chat referrals already account for 15%–20% of total referrals for some retailers, while some industry estimates suggest AI agents could handle as much as 25% of global e-commerce sales by 2030.

The direction is unmistakable:

The consumer may increasingly tell the machine what they want—and let the machine figure out what to buy.

 

Then comes Meta's Muse problem

Meta says Muse is a personal AI agent designed to actually perform tasks, including opening browsers, filling out forms and completing purchases with user approval.

But here is the important distinction.

Meta says Muse does not share a person's conversations or data in its virtual machine with Meta's advertising systems.

That doesn't mean advertising disappears from the economy.

It means something potentially more important:

The agent's purchasing decision is not necessarily another conventional advertising impression.

That creates a strategic question for retail media.

Suppose a consumer tells Muse:

“Order chicken breasts, vegetables and something easy for dinner tonight for under $25.”

The old advertising model says:

Show Tyson.

The agentic-commerce model says:

Find the product that best satisfies the consumer's criteria.

Those are not necessarily the same thing.

 

So why would Tyson advertise on Kroger?

This is where the industry needs to separate retail media from AI-mediated commerce.

Tyson might still have excellent reasons to advertise on a retailer's platform.

The advertising can:

1.       Reach human shoppers before they delegate the purchase.

2.       Build brand awareness.

3.       Influence a consumer's future preferences.

4.       Promote a new product.

5.       Drive trial.

6.       Communicate price or promotion.

7.       Reach shoppers inside a retailer's physical and digital ecosystem.

8.       Produce measurable sales attribution.

But if an AI agent increasingly determines which chicken goes into the basket, the value proposition of the ad changes.

The question becomes:

Is Tyson paying Kroger to persuade the consumer—or paying Kroger to make Tyson more visible and machine-readable inside an increasingly algorithmic marketplace?

That is a much bigger strategic question.

 

What about Tastykake on Wawa?

The same issue appears in convenience stores.

Wawa has enormous foodservice traffic, beverage traffic and impulse-purchase occasions.

A traditional consumer sees:

Tastykake → display → recognition → craving → purchase.

That is classic food marketing.

But an AI agent doesn't walk through the Wawa store experiencing the smell of coffee, seeing the pastry case and remembering childhood.

The agent receives a request.

“Get me breakfast and a snack.”

Now the competitive battlefield can become:

price + availability + preference + nutrition + convenience + previous purchase behavior + product data.

That doesn't eliminate the value of the Tastykake brand.

It potentially changes where the brand must establish its value.

 

And Budweiser on TGI Fridays?

This example illustrates another important distinction.

Restaurants are not merely grocery shelves.

A restaurant has:

·       atmosphere

·       social interaction

·       menu engineering

·       server recommendations

·       food photography

·       smell

·       presentation

·       entertainment

·       occasions

·       group behavior

·       impulse purchases

A Budweiser sponsorship, promotion or menu placement at a restaurant can influence a social occasion, not simply a product search.

That makes restaurants potentially different from grocery retail.

But even restaurants are moving toward digital ordering, loyalty, delivery, mobile ordering and AI-assisted discovery.

DoorDash's advertising expansion demonstrates how foodservice platforms are already becoming media businesses as well as transaction businesses.

The restaurant of the future may therefore have two customers:

the human customer—and the algorithm that helps determine what the human orders.

 

The biggest mistake would be killing advertising too early

Here is where the Grocerant Guru® sees an important distinction.

AI does not make advertising irrelevant.

It makes bad advertising less relevant.

There is a difference.

If the consumer says:

“I want something inexpensive, high-protein, ready in 10 minutes and under 600 calories.”

the food company that has the best machine-readable product information, price/value proposition, availability, nutrition information and verified attributes may have an advantage.

Deloitte calls this competition for the “algorithmic shelf.”

That means the next generation of food marketing may require two strategies:

Marketing to people

Brand + emotion + craving + experience + value

and

Marketing to machines

Data + attributes + price + availability + relevance + proof

The winning food marketer may need both.

 

Retail media must prove what it is actually selling

Retailers should be asking a harder question than:

“How much advertising revenue can we generate?”

They should ask:

“What decision are we influencing?”

That distinction becomes critical.

If an advertisement appears on Kroger's website and a human sees it, that is one type of value.

If an AI agent bypasses the ad and directly creates a basket from inventory, the retailer may still generate transaction revenue—but the advertising impression becomes less important.

And that could eventually put pressure on retail-media economics.

Instacart itself is already preparing for this transition. Its three-layer AI strategy includes:

·       Clementine, its native AI shopping experience;

·       Cart Assistant, which retailers can put on their own websites and apps;

·       Connectors, which allow outside AI platforms to access Instacart's grocery infrastructure.

At the same time, Instacart is aggressively expanding its advertising ecosystem. It reported more than $1 billion in advertising and other revenue in 2025.

That juxtaposition is fascinating:

The company is simultaneously building the advertising business and the technology that could help consumers shop without traditional browsing.

That is not necessarily a contradiction.

It may be the beginning of the next retail model.

 

The Grocerant opportunity: Stop thinking in silos

The Grocerant Niche has always challenged the industry's artificial boundaries.

Consumers don't necessarily think:

Grocery.

Restaurant.

C-store.

Delivery.

Retail media.

They think:

“What's for dinner?”

And increasingly they may tell an AI agent:

“Figure it out.”

That creates a massive opportunity for grocery stores, C-stores and restaurants.

The future competitive advantage may not simply be having the biggest advertising network.

It may be having the best answer to the consumer's prompt.

That means a grocery store needs to know:

·       what is available right now;

·       what is fresh;

·       what is ready-to-eat;

·       what is heat-and-eat;

·       what can be bundled;

·       what is on promotion;

·       what fits the consumer's budget;

·       what complements another product;

·       what can be delivered;

·       and what can solve the consumer's meal problem.

That's Grocerant thinking.

 

Mix-and-Match Meal Component Bundling becomes AI-ready

This is where Mix-and-Match Meal Component Bundling becomes even more important.

A human might see:

rotisserie chicken + salad + bread + dessert

and construct a meal.

An AI agent can potentially do the same thing—provided the retailer's data infrastructure understands the relationships among those products.

That creates an interesting shift:

Yesterday's merchandising strategy becomes tomorrow's machine-readable meal architecture.

The retailer isn't simply selling four products.

It is selling:

Dinner.

And that is a much more valuable proposition.

 

The new food advertising equation

The old equation was essentially:

Attention → Advertising → Desire → Purchase

The emerging equation could be:

Intent → AI interpretation → Product comparison → Recommendation → Purchase

Advertising doesn't necessarily disappear.

But its location—and its job—changes.

The food marketer may increasingly have to influence the inputs that AI uses to make decisions, while simultaneously maintaining enough human brand equity that consumers will recognize, request or approve the recommendation.

That's a very different marketing challenge.

And it could make product data, value and availability the new hand-held marketing.

 

Three Insights from the Grocerant Guru®

1. The advertising impression is not the purchase decision.

Retailers have spent years monetizing the consumer's attention. AI agents may increasingly monetize the consumer's intent instead.

The next retail-media question isn't simply “Did someone see the ad?”

It is:

“Did the product get selected?”

2. The algorithmic shelf is becoming as important as the physical shelf.

Tyson can win the physical shelf with packaging, placement and promotion.

Tastykake can win the impulse occasion with visibility.

Budweiser can win a social occasion with restaurant merchandising.

But the AI shelf requires something different:

relevance, value, availability and verifiable product information.

The brands that prepare their data for AI may discover that machine visibility becomes a new form of shelf placement.

3. The Grocerant winner will sell the solution—not the silo.

The consumer doesn't care whether dinner originated in the grocery store, C-store, restaurant, deli or delivery platform.

The consumer wants Price + Quality + Social + Portability = Value.

And increasingly, the consumer may simply tell an AI:

“What's the best dinner for me tonight?”

The retailer, restaurant, C-store and food brand that can provide the best answer—not merely the loudest advertisement—may be building the next generation of food marketing.

That's the Grocerant Guru® view: In an AI-powered food world, the most valuable media may not be the advertisement the consumer sees. It may be the product the algorithm chooses.

Tap into the Foodservice Solutions® team for greater understanding of New Electricity or for a Grocerant Program Assessment, Grocerant ScoreCard, or for product positioning or placement assistance, or call our Grocerant Guru®.  Since 1991 www.FoodserviceSolutions.us  of Tacoma, WA has been the global leader in the Grocerant niche. Contact: Steve@FoodserviceSolutions.us or 253-759-7869

 

Sources: GeekWire, Meta, Instacart, Circana and Deloitte, with 2022–2026 data and reporting cited above.