AI’s influence on snack choices explained
- Conagra found snack-related questions in its analysis of AI queries, including requests for products with specific nutritional benefits.
- Shoppers are using AI to research and compare groceries at home and in store, with assistants increasingly able to connect suggestions to online baskets.
- Whether an AI snack recommendation leads to a purchase remains unclear, making accurate product information and availability especially valuable.
‘Find me a crunchy snack with at least 10g of protein that I can buy nearby.’ Evidence suggests a growing number of shoppers are putting questions like this to AI assistants while planning a grocery order or standing in a supermarket aisle. Others might ask for a nut-free option for a child or snacks for a gathering that won’t break the budget. They can describe what they need instead of searching a retailer’s website one product at a time.
Conagra Brands says conversations like these are becoming part of how Americans choose snacks. Its latest Future of Snacking report combines an analysis of more than 53 million shopping transactions across 17,000 products with retail sales, search activity, digital behaviour and AI conversations. The Slim Jim maker argues that shoppers are increasingly looking for snacks that serve a purpose, including protein, energy or digestive health.
“Snacks are playing a bigger and more personal role in consumers’ lives, whether someone wants a boost of protein, lasting energy, or a moment of comforting indulgence,” says Bob Nolan, Conagra’s senior VP of growth science.
Conagra’s public findings, however, don’t say how many people asked AI about snacks, how often they did so or whether a recommendation led to a purchase. Its transaction data establish what sold, while the AI conversations may offer clues about what shoppers wanted. Without a link between the two, the report doesn’t establish whether AI contributed to the growth in functional snack sales. But manufacturers still have good reason to examine when shoppers turn to AI and what happens after they receive a suggestion.
When does a snack become a question?
Shoppers may ask for help when several requirements need to be met at once. Conagra’s report gives ‘What is the best low-calorie protein snack?’ as an example drawn from its analysis of AI queries. ‘Show me high protein snacks’ is also among the suggested requests in Instacart’s guidance for Clementine, the AI assistant it introduced across its North American marketplace in September. Shoppers can ask it for products, recipes or a list, then add suggested items to a cart at their chosen store.
Instacart also describes a request for everything needed for a football watch party, which could bring snacks into the conversation without the shopper searching for the category directly. A dietary restriction, a nutrition target, a budget and an occasion can all shape the answer. Someone searching for ‘protein snacks’ on a retailer’s website may still have to inspect dozens of products. An AI assistant can take a fuller request – something savoury, portable and within a particular price range, for example – and use those details to narrow the choice.
McKinsey’s 2026 European grocery research puts some scale around this behaviour. About 12% of consumers had used AI during their grocery shopping journey. Among those users, 40% had used it to create meal plans, 37% to research specific products and 36% to find the best prices or recipes suited to dietary preferences. Meal planning is the leading use in that group, but the product research and price checks show how AI can enter decisions closer to the shelf.
In the US, Acosta Group found that 34% of shoppers used AI tools to shop, chiefly to save money or time, compare products and research purchases. Among Gen Z respondents who used AI tools, 55% said they used them while shopping in a store. The survey, conducted with 1,271 US shoppers in May, suggests that AI advice isn’t confined to planning at home. Some shoppers are consulting it while looking at the products they could buy there and then.
A request can also move between platforms. A shopper might begin with a general chatbot, use an assistant within a grocery app while building an order or check a suggestion on their phone beside the shelf. Instacart says grocery conversations that start in ChatGPT, Claude, Gemini or Google’s AI Mode can be completed through its ordering service. That gives a recommendation a potential route into the basket, provided the suggested product is available at the shopper’s chosen store.
Conagra’s report puts that route in a larger commercial context. E-commerce accounts for 13% of US snack dollar sales, up two percentage points year on year, while online snack baskets are 36% higher in value than in-store purchases. If an AI suggestion sends a shopper to an online store, the product still has to be listed accurately, in stock and easy to add to the basket.
Discovery is easier to prove than a sale
Conagra has a strong commercial reason to pay attention to these requests. Its report puts annual US sales of snacks with targeted benefits, including protein, energy, digestion and hydration, at $19bn, up 12.7% on the previous year. High-protein snacks alone generated $12.1bn. Shoppers’ interest in these attributes is apparent in sales as well as in the kinds of questions an assistant is built to answer.
According to Nolan, cookies and salty snacks have come under pressure as shoppers seek products that meet more specific needs. “The emergence of these more nutrient-dense snacks and snacking for purpose have started really growing and accelerating,” he notes. AI could make it easier to identify products that fit those purposes, particularly when shoppers are weighing several attributes at once.
Acosta found that half of Gen Z and millennial shoppers who used AI tools said the technology had influenced a buying decision in the previous three months. Most commonly, it helped them choose between products. A snack brand could therefore benefit from appearing among the suggestions even if the shopper checks the pack and compares prices before deciding. Acosta also found that 40% of shoppers influenced by AI had tried a new brand, suggesting that recommendations can introduce products they might otherwise miss.
Shoppers appear less willing to let an assistant make the purchase for them. In Acosta’s study, only 22% of people using generative AI for shopping trusted an agent to make purchasing decisions on their behalf. Gartner likewise found US consumers more receptive to help narrowing choices than to AI deciding what to buy. Among people who had used AI during a recent purchase, 54% said they’d had to check the accuracy of all the information it supplied.
Checking the answer can be especially consequential with food. Instacart warns that its assistant’s information on ingredients, allergens and nutrition may be incomplete or out of date, while prices and availability can vary by store and location. A ‘high protein’ recommendation offers little help if the product is out of stock or its nutritional details are wrong. An incorrect allergen suggestion carries a more serious risk.
What should snack makers take from it?
Manufacturers can start with the information an assistant needs to assess a product against a shopper’s request. If someone asks for a filling afternoon snack, an option that meets a dietary restriction or food for a group, accurate ingredients, nutrition, allergens, pack sizes and availability all help determine what fits. Those details also give the shopper something reliable to check before buying.
Brands can then examine what shoppers actually see. When an assistant is asked about a relevant occasion, does it suggest the product? Does it describe the product correctly, and can the shopper buy it at the stated price from a nearby retailer? Repeating those checks across different requests could expose gaps in product information or availability that a conventional shelf audit wouldn’t catch.
So, are consumers really asking AI what snacks to eat? Some are, and Instacart explicitly suggests ‘Show me high protein snacks’ is a regular request for its grocery assistant. Broader surveys show shoppers using AI to research groceries, compare products and resolve choices both at home and in store. What remains unproven is how often people ask a snack-specific question and how much influence the answer has on sales.
Conagra has identified a plausible new point of influence; the next useful evidence would connect the question a shopper asks to the snack they actually buy.
