Nearly four in 10 Canadian shoppers have used artificial intelligence to help them shop over the past year, as AI begins to play a larger role in how consumers discover, research and compare products.
Salesforce research found that 39 per cent of Canadian shoppers have used an AI tool for shopping during the past 12 months, according to Caila Schwartz, Director of Industry Insights at Salesforce. Roughly 11 to 12 per cent now say they start their shopping journey with AI, up from single-digit levels last year.
“I would argue last year it was a niche type of behaviour, and in 2026 AI search and discovery is becoming a much more mainstream shopping behaviour for the Canadian shopper,” Schwartz told Retail Insider.
The shift is creating new questions for retailers around product visibility, data quality and marketing measurement as more of the shopping journey takes place inside AI conversations. Salesforce’s State of Commerce research found that 89 per cent of Canadian commerce professionals believe AI and large language models will be essential to product discovery within the next year.
Canadians Are Using AI to Find Value
Price comparison and deal hunting are currently the leading shopping uses for AI among Canadian consumers, according to Schwartz.
Consumers are also using the technology to conduct research quickly and receive product recommendations, reducing the need to move between multiple websites and search results.
“For the Canadian consumer, the number one reported behaviour is to compare prices and find deals,” Schwartz said. “The second is being able to do all of that research in one go very, very quickly. It does all the research for them. And then the third is that it gives them the best product recommendations.”
The way consumers use AI also varies by income. Schwartz said higher-income shoppers tend to place greater emphasis on personalization, while other consumers are leaning more heavily on AI to find deals, coupons and conduct price comparisons.
Traditional search remains important, and Schwartz cautioned against interpreting the growth of AI discovery as a direct transfer of market share away from search engines. Salesforce is seeing fewer consumers report using some traditional discovery channels and more report using emerging ones, but its research does not measure the entire discovery market as a fixed pool of traffic being redistributed between channels.
The use cases are also different. A consumer who already knows exactly what they want may still favour a conventional search, while AI can be more useful when a shopper is comparing options, seeking inspiration or working through a more complex purchase.
Retailers Face a New Visibility Challenge
The growth of AI shopping raises an increasingly important question for retailers: when a shopper asks an AI assistant what to buy, what determines which products and brands appear in the answer?
There is no settled formula yet.
“I think that is what everyone is still trying to figure out,” Schwartz said.
Salesforce believes data quality will play a central role. Schwartz pointed to clean and consistent product information, accurate inventory and pricing, and direct catalogue integrations as areas retailers should be considering as AI-driven discovery develops.
“Ultimately, it comes down to data,” she said. “Having a really sound data strategy, having clean and consistent data, trusted data across your product detail page, across your entire ecosystem.”
Direct catalogue integrations could help AI systems work with current inventory and pricing rather than recommending merchandise that is out of stock or displaying outdated information. Schwartz said Salesforce believes these tactics can also improve how retailers and their products appear within AI search results.
Retailers are consequently beginning to think about visibility beyond traditional search engine optimization. Concepts including generative engine optimization and answer engine optimization — commonly shortened to GEO and AEO — are emerging as companies consider how their information should be structured for AI-driven discovery.
Schwartz described it as, in some respects, “the new version of SEO.”
The rules remain fluid. Retailers are entering an environment where the factors influencing AI-generated recommendations are newer and less established than the practices that evolved around conventional search.
Canadian Retailers Are Moving Quickly
Canadian commerce professionals appear well aware of the change. Salesforce found that 89 per cent believe AI and large language models will be essential to product discovery within the next year. About 34 per cent are already using agentic AI, while another 33 per cent plan to use it within the next six months.
“They are seeing this already today,” Schwartz said. “They are prioritizing AI adoption.”
The larger challenge may be what sits underneath those systems.
Only 30 per cent of Canadian retailers surveyed by Salesforce said their customer data is fully unified across areas including sales, marketing, commerce and service. That fragmentation could make it harder to deliver consistent, personalized experiences as retailers increase their use of AI.
Schwartz said Canadian retailers that have successfully unified their data reported better sales outcomes, stronger AI automation for their teams and improved customer retention.
“Having a sound data strategy is really going to be important,” she said.
Smaller Retailers May Have an Opening
The transition to AI-driven product discovery does not necessarily favour the largest retailers, according to Schwartz. She described the current environment as a relatively even playing field and said smaller brands and retailers have an opportunity to establish themselves while AI commerce remains in its early stages.
Large organizations may have greater technology resources, but they can also face more complicated data environments. Multiple systems and data silos can make it difficult to create the unified information needed to support AI applications.
“Those things can be hard for larger organizations because they have more systems, more data silos that they have to unify,” Schwartz said.
She believes the technology could act as an equalizer across parts of the industry, particularly for smaller companies able to maintain clean product information and a coherent data strategy.
Whether that ultimately translates into greater visibility for smaller retailers will depend on how AI shopping tools and recommendation systems continue to develop.
Retail Marketing Has a Measurement Problem
The movement of product discovery into AI conversations is also creating a challenge for marketing attribution. A shopper can now conduct substantial research inside an AI assistant before visiting a retailer. Recommendations, comparisons and part of the purchasing decision can happen without generating the website interactions marketers have traditionally used to understand the customer journey.
Schwartz said there is not yet a clear answer for measuring that activity.
“If the discovery is staying in the conversation, we don’t have visibility into any of that,” she said. “In terms of measurement, I don’t think we have an answer for that yet.”
The issue could become increasingly important as AI takes a larger role at the beginning of the shopping journey. Search referrals, website sessions and conventional digital attribution may capture only part of the process if consumers arrive after completing much of their research elsewhere.
For retailers, Schwartz said the immediate priority is controlling what they can: the quality and structure of the data that determines how they and their products show up in these emerging channels.
Much of the Coming AI Traffic Will Be Bots
Retailers will also have to manage another consequence of AI adoption: an increasing share of website traffic may not be human.
Salesforce predicts that 20 per cent of holiday e-commerce traffic in 2026 will originate from AI agents. Schwartz clarified that the majority of the traffic represented in that prediction is expected to come from bots rather than consumers directly arriving from AI conversations.
“The majority of that is bots,” she said. “A lot of that is non-human.”
AI systems use crawlers to visit retail websites, access product pages and collect information for indexing or to bring information back into conversations with shoppers.
That creates a trade-off for retailers. They want their products accessible to AI systems because those systems may increasingly influence purchasing decisions, but large volumes of automated traffic can put additional demands on websites, particularly during peak shopping periods.
Schwartz said retailers may have to make decisions about which systems they allow to crawl their websites during periods of particularly heavy traffic. She also pointed to direct catalogue integrations as one way to reduce reliance on crawlers while providing AI systems with more structured and current product information.
The industry is already experimenting with other approaches.
“We might even see some websites that are built just for bots and not for humans,” Schwartz said.
The idea points to a potentially significant change in e-commerce infrastructure, where retailers could increasingly provide information to automated systems through channels designed specifically for machines rather than relying exclusively on the same customer-facing websites used by human shoppers.
From Discovery to Purchasing
The next question is how far consumers will allow AI to go. Using an AI assistant to research products carries relatively little risk. Allowing one to select an item and complete a transaction requires a higher level of trust.
Schwartz believes the industry may be closer to that point than it appears, although she does not think third-party AI systems have yet become significant purchasing channels.
Social commerce provides an indication of how behaviour could evolve. Schwartz said consumers, particularly Gen Z and Millennials, have become increasingly comfortable completing purchases inside social applications rather than clicking through to a retailer’s website.
“I don’t think AI is there yet as a purchasing engine,” she said. “But I think what it does tell us is that consumers will eventually be comfortable purchasing in these tools.”
Retailers’ own AI agents could reach that point sooner. A shopper interacting with an AI assistant on a retailer’s website may already have a loyalty profile and saved payment information, which Schwartz said could make consumers more willing to complete a transaction within the conversation.
For now, third-party AI tools are primarily changing product discovery. Even at this stage, retailers are having to reconsider how products are presented online, how customer and catalogue data is managed, and how marketing effectiveness is measured.
With 39 per cent of Canadian shoppers already using AI as part of shopping and 89 per cent of Canadian commerce professionals expecting AI and large language models to become essential to product discovery, the shift has moved beyond an emerging consumer experiment.
For retailers, the challenge now is ensuring their products, pricing and availability can be accurately understood by the AI systems increasingly helping consumers decide what to buy.
