Like most large retailers, Canadian Tire has plowed a lot of resources and staff energy over the decades into the placement of products within its stores, codifying and refining crucial decisions: what should go where—and when—to maximize sales.

The maps showing those constantly shifting layouts are known as “planograms,” and in the past they’ve been an alchemy of technology and intuition, drawing on both employees’ institutional knowledge and the output of increasingly sophisticated software.

Now, the iconic retail giant hopes to take the next leap forward, using artificial intelligence to process the thousands of data points that go into the most profitable configuration. The project is codenamed “Tetris”—a nod to how these store maps solve the shelf-stocking puzzle—and it’s the biggest product to date from a Canadian Tire team dedicated to solving common retail problems using AI.

The leader of the unit believes Canadian Tire is the first retailer to successfully develop planograms at scale. “The solution doesn’t exist anywhere else,” said Cari Covent, whose title recently changed to head of AI and emerging technology to reflect the company’s ambition in the space.

Before Tetris, Canadian Tire’s planograms came in five standard sizes, each entailing a specific set of products. They relied on national forecasts to determine how much stock stores needed of certain items, and were updated infrequently. They couldn’t change to adapt to things like new packaging dimensions for products.

Covent and her team set out to create a program that could make store-specific planograms that better reflect the needs of customers in a given region. To do that, though, the system would have to process a multitude of factors the old methods could not.

With Tetris, managers can customize aisle dimensions to map what their actual space looks like, rather than relying on one of five pre-set options. They can customize what products they want to include, even adding discontinued items. The system factors in each store’s sales forecast to determine product stock, and regularly updates item data like package dimensions so managers can refresh their planograms as those inputs change.

Under the new system, stores of similar size in Toronto and Moose Jaw, Sask., would have tailor-made plans. The Moose Jaw location, for example, may allocate more space for wood pellets, as people in its area are more likely to use wood stoves than their Toronto counterparts.

Canadian Tire approached Scale AI, the federal government’s AI supercluster, to help fund its work on Tetris. Scale AI provided it with $5.6 million over three years and connected it with a startup that could help build the solution, Montreal-based Ivado Labs. (Ivado did not respond to a request for comment.)

On a project like Tetris, Canadian Tire’s partnership with the startup will evolve, said Covent, with in-house staff steadily assuming more responsibility as they learn. “Our team works alongside them up until a certain milestone and once we hit that milestone, we take on the bulk of that development,” she said.

The retailer rolled out Tetris to six stores in a pilot a couple of years ago. Since then, the system has been available to 244 Canadian Tire locations, which use it primarily to reset their merchandise as the seasons change. The company is expanding the roll-out to its remaining Canadian Tire stores, of which there were 421 at the end of last year, and looking at how it can expand it to its other banners, such as SportChek, Mark’s and Helly Hansen.

It’s too soon to say whether Tetris has boosted revenue, said Covent, but store managers using the new tool have reported significant time savings, with fewer worker hours spent planning, building and adjusting new displays.

Maxime Cohen, the Scale AI professor of retail and operations management at McGill University and a scientific adviser at Ivado Labs, was impressed. “As soon as you start using computer software and AI to guide those decisions, you’re always one step ahead of your competitors,” he said.

Scale AI looks to fund projects that help industries optimize their supply chains, and found “a huge interest in AI coming from retailers,” said CEO Julien Billot, adding the industry was the first to turn to AI to boost their businesses.

The attraction lies in the technology’s capacity to smooth out the sort of sprawling and complex operations typical in large-scale retail, said Billot. It’s better than conventional methods at predicting demand; forecasting inventory needs; optimizing supply chain operations; and enhancing logistics. Retailers can use it to decide what each item should be priced at every week, manage inventory and personalize product recommendations, added Cohen.

Scale AI has worked on projects in some of these areas with shoe retailer Aldo and grocery chain Save-On-Foods, among others. But Canadian Tire, whose initiative was one of the first Scale AI funded, has been ahead of the game, said Billot.

Indeed, the company was one of few Canadian retailers publicly discussing its AI plans before last November’s launch of ChatGPT, which sent interest in generative AI soaring. The retailer’s CEO was mentioning its Tetris plans on conference calls with analysts as early as 2020, transcripts from PitchBook show.

Covent started Canadian Tire’s AI unit back in September 2017, when the company wanted to automate certain repetitive manual processes in finance. Inside the company, the group is known by a couple of acronyms—CTX (the X stands for “exponential”) and DNA (“data into action”). Its mission is to help  Canadian Tire meet its strategic objectives through the use of AI, she said.

In addition to Tetris, Covent’s team helped create a chatbot called Copilot to assist customers shopping online. Part of a partnership with Microsoft announced in June, the system will answer questions through voice or text, Covent said, using conversational style and a customer’s preferred language. Canadian Tire plans to launch a beta version this year and make it available to more customers in the spring.

During the pandemic, the company used artificial intelligence to help predict what products customers would want before demand for them started to rise, Rex Lee, Canadian Tire’s chief information and technology officer, said in late September at Scale AI’s All In conference in Montreal. That helped it move items between stores from different banners so they were there for customers when demand peaked. “As others didn’t have product, we were there,” Lee said.

Covent’s team has also focused on automation, leading the retailer’s partnership with Vancouver-based Sanctuary AI, which has been developing a humanoid robot it believes could replace humans in many types of jobs. Sanctuary’s robot, Phoenix, has worked at a Mark’s and a SportChek store to test its ability to do retail-worker tasks in a real-life environment. 

Covent was interested in how Canadian Tire might use the robots in its retail stores and distribution centres, potentially offsetting the impacts of an anticipated global labour shortage. The team is pleased with Phoenix’s progress and is now assessing how best to put it to work in operations, she said, adding, “At this point, we’re looking at all options.”

Covent and her bosses believe strongly in AI’s transformative potential for Canadian Tire—an ambition represented by Copilot and the Microsoft partnership. “Exploring the use of generative AI and all of its capabilities at Canadian Tire is front and centre,” she said. The vision, as described by Lee, is sweeping. “Every single function, every single role, every operation in Canadian Tire will be changed by AI. I have no doubt about that,” he said. “I’m not saying tomorrow. But it will shift.”