Hitmetrix - User behavior analytics & recording

Personalization Helps Marmot Improve Conversion Rates

 

There’s nothing like an in-store experience. Customers can discover a vast array of new products, and sales representatives can make personalized recommendations to get them to buy. As the world becomes more digital, marketers are trying to reproduce these tailored experiences through recommendation algorithms. And although nothing can quite replicate face-to-face communication, brands like outdoor apparel and equipment company Marmot are giving it their all and leveraging customers’ behavioral data to create individualized experiences that drive conversion.

Embarking on a new adventure

Jeff Milbourn wanted to find a way to customize consumers’ interactions with Marmot.com and make their online experiences more relevant based on their interests and needs. So the director of e-commerce for Jarden Technical Apparel—the division of Jarden Corporation that houses Marmot—started looking for a personalization platform that could help deliver these tailored experiences and, ultimately, boost the brand’s engagement and conversion rates.

“We felt…that [it] would promote an experience of discovery that [would] impact user shopping experiences in a deep and meaningful way,” he says.

But finding the right platform provider was easier said than done. Many of the platforms that Milbourn considered relied on “overly large, broad” segmentation, he says, and could only guarantee single-digit conversion lift. So when he learned about personalized commerce solution provider Reflektion and its machine learning capabilities, he was intrigued; plus, the provider’s promise to generate at least a 10% increase in conversion rates only sweetened the deal.

To see if the platform was the real deal, Milbourn and his team implemented it in January and ran a month-long pilot program.

Utilizing the right gear

For the pilot program, Milbourn tested two solutions: Reflektion’s Predictive Product Merchandising tool and its Individualized Site Search capability.

The Predictive Product Merchandising tool leverages Web behavior, clothing categorization, and machine learning to deliver personalized recommendations. The more customers engage with Marmot and its products, the more refined the recommendations become.

On the product pages, for instance, there are two types of recommendation widgets: “Similar Items” and “You May Also Like.” As the name suggests, the “Similar Items” widget shows consumers products that are comparable to the one featured on the page based on its category of clothing (e.g.women’s hoodies, men’s vests). This offers alternatives in case the original product isn’t exactly what they’re looking for.

 

The “You May Also Like” tool addresses a different need. Based on consumers’ Web behaviors, it shows them items that are similar or complementary to the featured product and serves as an upsell opportunity. In addition to appearing on the Marmot product pages, the tool appears on the checkout page and shows consumers items similar to the ones they have in their carts.

 

The company also has a “Popular This Season” section on its homepage. If customers are new to the site, then Marmot will show them items that are trending. But if they spend time clicking on products and engaging with the widgets, then the items featured in the section will reflect their behaviors.

 

As for the Individualized Site Search solution, consumers can click in the search box to see a visual representation of items that are trending. But when they start to type in a search query, the solution will predict what they’re looking for and show them items that match.

 

Milbourn says these platform solutions allow consumers to experience more pertinent experiences and find the products that they want more quickly.

“[The platform is] able to capture a user’s true interest or intent and then serve up content that’s going to be relevant to them,” he says. “So, it helps them navigate a portfolio of products in an easy way, but it also promotes the discovery of the product line based on the interest and intent expressed through click behavior. It resonates quite a bit with individuals when you serve up content that they’re interested in.”

Setting up camp

To track customers’ Web behavior, Milbourn had to provide Reflektion with Marmot’s product feed and install a tracking pixel on the brand’s site. The Reflektion platform also stores a user identifier via a cookie on the customer’s device. As a result, all of the Web behavior is tied to that identifier and that device. So if a customer starts browsing Marmot.com on his tablet and then continues his research on his desktop, his recommendations will not reflect his mobile activity.

As for measuring the success of the pilot program, Milbourn says that Reflektion conducted an A/B test and only showed the personalized experiences to half of its site visitors. This test lasted for about four weeks. Marmot also linked the Reflektion platform to Google Analytics to compare engagement and conversion results for both groups.

Climbing to the top

During the pilot program, Marmot experienced a 13% increase in conversion rate among customers exposed to the Reflektion experience.  Milbourn also says that the brand saw a 27% lift in average order value among people who engaged with the recommendation widgets on the product pages, versus those who just viewed them. Plus, it experienced a 10% decrease in exit rates and 13% decrease in bounce rates on its product pages.

For Milbourn, the results of this pilot only solidify personalization’s role in the marketing sphere. As he simply puts it, “Marketing to individuals—one-to-one personalization—is the future.”

Update 5/4/16: Further clarification regarding the data used to generate recommendations within the “Similar Items” and “You May Also Like” tools was added.

Total
0
Shares
Related Posts