For Black Friday, Canadian Tire Corporation, a mass retailer that sells everything from soda streams to hockey sticks, had an ambitious target: beating Amazon at the online retail battle they’ve owned for more than a decade. Quite a challenge since Canadian Tire invests nothing close to the billions Amazon invests online.

Together, Amazon and Canadian Tire sell millions of different products via their websites. In this cluttered environment, featuring the right “star products” on their ads and betting on the right categories is critical to success. In retail, these “star products” are usually selected by merchandising experts months in advance, missing out on the opportunity to react to changes in consumers’ product preferences.

In the days leading up to Black Friday, we detected a new analytic trend: the average number of items left abandoned in Canadian Tire’s website carts increased by the hour. We called this new trend “online window shopping”: placing items temporarily in carts, consumers look for similar options on competitors’ website. Consumers were prepping up for Black Friday, ready to buy as soon as deals were on!

We quickly realized that these temporary carts were a golden mine of insights. They told us, in real time, which items, out of the hundreds of thousands sold by Canadian Tire, could be used as “online door crashers”, also revealing what products to leverage to win against Amazon.

Leveraging “online window shopping” brought new visitors who generated either e-comm sales…or abandoned items in their carts, providing more real-time data from which we selected more star products. An infinite loop of insights and sales.

A live data stream was quickly set up to generate insights from the online window shopping phenomenon, and fuel the entire campaign, from product selection to creative development, in real-time.

This new approach was everything but a mere re-targeting campaign. We did not use carts data to re-target a small pool of visitors who had abandoned them. To win Black Friday, a 24-hour sale event, the insights generated created massive reach in a record time. Items that we found in abandoned carts were quickly featured in our mass campaign, reaching 30% of Canadians consumers in only a few hours.

We also used this live data feed to select which Search categories to bid on. Canadian Tire simply did not have the means to win the bidding war against Amazon across all categories. But thanks to the rich insights found in the carts, we could win all the right categories.

The star products to be advertised and key categories to bid on were selected in real time, using data from abandoned carts, and drove the media execution. The top 1,000 products left in carts URLs and associated SKUs were fully leveraged to create new, dynamic campaigns in only a few hours.

Canadian Tire ads engaged a wide audience of high potential consumers on a variety of digital touch points, from YouTube to Google search to Instagram. We relentlessly optimized the campaign towards the most performant placements, formats, creatives.

The data trail left by the “online window shopping” trend drove the optimal Search strategy. And Amazon was beat in key categories, with a fraction of their overall investment. There were over 78 M bid changes over the short campaign duration.

Finally, thanks to custom-built scripts, abandoned carts data was crossed with inventory data and we stopped advertising products as

Year over year sales went up by 54%. See confidential information for more details.