AI and e-commerce: What retailers need to know about Canadians’ changing shopping habits

Artificial intelligence is no longer just a topic of innovation or strategic planning. It is already beginning to influence how consumers search for, compare, and choose products online.

This is one of the key findings from the fourth edition of the eCommerce Trends Report, conducted by Novatize in collaboration with Léger. This year’s study highlights an important shift: AI is gradually becoming part of Canadians’ shopping habits, particularly when it addresses a concrete need.

According to the 2026 results, 52% of Canadians have already used an AI tool in an online shopping context, compared to 39% in 2025. Openness toward these tools is also growing, with 63% of respondents now saying they are likely to use AI as part of their shopping journey.

“AI in shopping is no longer a thing of the future: more than half of Canadians have already used it in an online shopping context.” — Thavy Khamtan, Novatize

For retailers, the message is clear. The question is no longer whether AI will impact the retail sector, but rather where it can create meaningful value for customers.

Consumers are not using AI simply because it is trending. They are using it when it helps them make better purchasing decisions.

The study identifies product or brand comparison as the leading use case, with 44% of Canadians saying they are likely to use AI to compare products, understand the differences between similar options, or evaluate one brand against another.

This finding is particularly important for retailers. It shows that AI is perceived as a decision-making tool, not just a conversational one. Customers are looking to save time, reduce uncertainty, and better understand what sets one option apart from another.

In a context where online catalogs can contain hundreds or even thousands of products, this expectation becomes increasingly important. The broader the offering, the more customers need support to navigate, filter, compare, and choose.

Not all product categories are affected in the same way. The results show that interest in AI is stronger in categories where the purchasing decision is more complex, technical, or high-stakes.

Electronics leads the way, with 28% of Canadians saying they are likely to use an AI tool for this category. Appliances follow at 25%, while home and décor and health and wellness are tied at 22%.

These results are not surprising, given that these are categories where consumers often need to compare multiple features, understand technical differences, assess compatibility with their needs, or make a more expensive purchase.

For retailers, this means AI should not be deployed uniformly across every category. Instead, priority should be given to areas where it can genuinely help customers make a more informed decision.

For example, an appliance retailer could use AI to help customers compare models based on available space, energy consumption, features, budget, and delivery times. A home improvement retailer could use it to guide product selection based on the type of project, the customer’s level of expertise, or compatibility with other materials.

In these cases, AI becomes a tool for clarification. It reduces friction at the point when customers need the most help.

The growing adoption of AI does not mean that every retailer needs to quickly add an intelligent assistant to their website. A poorly targeted integration can even add complexity rather than simplify the customer experience.

Instead, the first step is to identify the moments in the customer journey where customers need the most support.

1. Prioritize complex decisions

Retailers should start with the categories where customers hesitate the most, compare options more extensively, or contact customer service more often before making a purchase.

A few useful questions to ask:

• Which products generate the most questions before purchase?
• Which categories have high return rates?
• Where do customers drop off in their journey?
• Which purchases require the most comparison or explanation?
• What information is often missing or difficult to understand?

These areas often represent the best starting points for testing AI use cases.

AI depends heavily on the quality of the data available. If product pages are incomplete, inconsistent, or poorly structured, AI tools will have more difficulty interpreting the product offering, recognizing products, and providing relevant recommendations.

Retailers should therefore review the quality of their product information: titles, descriptions, attributes, variants, compatibility, dimensions, availability, return policies, delivery times, and customer reviews. To make comparisons possible across different environments, it is also becoming essential to use universal identifiers such as SKUs, GTINs, UPCs, or EANs. These identifiers make it easier to reliably recognize a product, even when it is presented across multiple platforms, channels, or catalogues.

Beyond the content visible to customers, the technical structure of this information is becoming just as important. The use of semantic HTML, standardized attributes, and structured data—particularly JSON-LD based on Schema.org standards—helps search engines, AI assistants, and conversational agents understand products more accurately. It also makes product attributes more readable in the formats expected by new protocols related to agentic commerce, such as UCP and ACP.

Without this structure, AI tools may make assumptions, misinterpret certain information, or simply overlook a product. Good product data is therefore no longer just about improving the experience on a website. It is becoming essential for products to be understood, compared, and recommended in shopping journeys increasingly influenced by AI.

2. Improve the quality and structure of product information

AI depends heavily on the quality of the data available. If product pages are incomplete, inconsistent, or poorly structured, AI tools will have more difficulty interpreting the product offering, recognizing products, and providing relevant recommendations.

Retailers should therefore review the quality of their product information: titles, descriptions, attributes, variants, compatibility, dimensions, availability, return policies, delivery times, and customer reviews. To make comparisons possible across different environments, it is also becoming essential to use universal identifiers such as SKUs, GTINs, UPCs, or EANs. These identifiers make it easier to reliably recognize a product, even when it is presented across multiple platforms, channels, or catalogues.

Beyond the content visible to customers, the technical structure of this information is becoming just as important. The use of semantic HTML, standardized attributes, and structured data—particularly JSON-LD based on Schema.org standards—helps search engines, AI assistants, and conversational agents understand products more accurately. It also makes product attributes more readable in the formats expected by new protocols related to agentic commerce, such as UCP and ACP.

Without this structure, AI tools may make assumptions, misinterpret certain information, or simply overlook a product. Good product data is therefore no longer just about improving the experience on a website. It is becoming essential for products to be understood, compared, and recommended in shopping journeys increasingly influenced by AI.

3. Make comparisons more useful

Since comparison is the strongest use case, retailers should structure their content around the criteria that actually influence purchasing decisions.

A good comparison tool should not simply display products side by side. It should help customers understand what really differs from one option to another: performance, recommended use, compatibility, durability, maintenance, total cost, availability, warranty, or level of service.

To achieve this, information needs to be standardized and comparable from one product to another. For example, two products should not describe the same feature in three different ways. Attributes should be consistent, easy to understand, and presented in formats that are easy to interpret, both for customers and AI tools.

The goal is therefore not simply to provide more information, but to make that information clearer, more reliable, and easier to use. This structure allows customers, search engines, and AI agents to compare the right information at the right time and support more informed purchasing decisions.

4. Summarize customer reviews

Customer reviews have become an essential source of information, but they can be difficult to navigate when there is a large volume of them. AI can help summarize recurring themes such as quality, sizing, comfort, installation, durability, noise, performance, after-sales service, or value for money.

For retailers, this is an opportunity to help customers understand a product’s strengths and limitations more quickly. It can also strengthen trust, provided that summaries remain transparent, nuanced, and based on genuine customer reviews.

5. Keep a “human in the loop”

AI can simplify many steps in the customer journey, but it does not completely replace the role of customer service representatives or advisors.

For complex purchases, sensitive situations, or cases where trust is essential, human support remains important. Retailers should therefore design hybrid experiences: AI can answer simple questions quickly, narrow down options, or summarize information, while people can support customers through more complex decisions.

The growing integration of AI into the customer journey is gradually transforming the role of product data. It is no longer used solely to power a transactional website, catalogue, or advertising campaign. It now influences discoverability, recommendations, and a brand’s ability to be accurately interpreted by automated systems.

In an environment where customers use tools such as ChatGPT, Claude, Gemini, or Copilot to compare products, retailers need to ensure that their offerings are easy to understand for both humans and AI agents.

This requires strong foundations: a clean catalogue, consistent data, useful content, clear rules, and reliable operational information.

Poor-quality data can lead to inaccurate recommendations, reduced visibility, or a disappointing customer experience. Conversely, well-structured data can become a direct competitive advantage.

AI cannot create a good customer experience if it relies on fragmented information. For a tool to recommend a product, confirm its availability, take the purchasing channel into account, or guide a customer toward the right option, data needs to flow across the company’s different systems.

This is where unified commerce becomes an important foundation. Retailers that better connect their product data, inventory, pricing, promotions, orders, customer service, and sales channels are better positioned to deliver AI experiences that are genuinely useful.

For example, an AI recommendation is much more valuable when it takes in-store availability, delivery times, return policies, customer preferences, or purchase history into account. Conversely, an AI experience that is poorly connected to operations can create confusion, recommend unavailable products, or provide inconsistent answers.

For retailers, the challenge is therefore not simply to add an AI layer to the customer journey. They first need to ensure that the foundations of unified commerce are in place so that the experience is reliable, consistent, and genuinely useful, regardless of the channel the customer uses.

This perspective also echoes the takeaways shared at our post-NRF event held in Montreal last January. In a previous article published on the CQCD blog, we already explored the rise of agentic commerce and UCP, and what these new standards could concretely mean for retailers. The findings from the Novatize and Léger study reinforce the same idea: to prepare for customer journeys increasingly influenced by AI, retailers first need to focus on solid foundations, reliable data, and an architecture capable of connecting their operations.

The 2026 results show that AI is a real trend, but they also call for a measured approach. Not every AI investment is automatically relevant. Consumers are adopting AI when it removes a real point of friction in the customer journey, not when it adds a layer of technology without a clear value.

For retailers, the right question is therefore not, “What AI technology should we deploy?” Rather, it is, “What purchasing decision do we want to help our customer make?”

Starting with that question makes it easier to identify the right use cases.

The 2026 eCommerce Trends Report highlights the growing role of artificial intelligence in Canadians’ shopping habits. For retailers, the challenge is no longer simply to adopt new technologies, but to ensure their organizations are ready to adapt to these evolving customer journeys.

This transformation relies on essential foundations: high-quality data, well-structured content, and interconnected systems. Together, these enable retailers to become a reliable source of information—not only for their customers, but also for the AI agents that are increasingly influencing purchasing decisions.

Get a strategic assessment of your readiness for agentic commerce. Novatize offers an assessment to evaluate your current presence, the quality of your data and content, and the structure of your commerce ecosystem, helping you better prepare for the era of agentic commerce.

Discover Novatize’s strategic agentic commerce assessment.

Download the full report. To explore the full findings from the 2026 eCommerce Trends Report: AI and eCommerce — What Canadians Are Actually Adopting in 2026, read the full report from Novatize, developed in collaboration with Léger.