Seasonal Help

There’s no one better to give advice to would-be adventurers than those who are also passionate about skiing, camping or hiking. That’s why Sun & Ski Sports employs sales associates who also happen to be experts in adventure.

It works easily enough in the retailer’s 17 stores, but transferring that same level of knowledge to an online shopping experience has been an adventure all its own. Because of the seasonality of its product line, it was extremely difficult to offer suggestions based on previous purchases. Seasonality also can alter a shopper’s level of interest. A passionate cyclist may be willing to spend top dollar for the best equipment in his area of interest, but not on a tent for a weekend outing.

Sun &SkiSun & Ski tapped Baynote Product Recommendations to optimize the recommendation end of the site, which boasts 15,000 SKUs from 100 brands. Baynote collects data on similar shoppers and then tailors the recommendations based on their search behaviors.

“Allowing the site to dynamically and automatically adapt its links and structure is a much more scalable approach,” says Scott Brave, Baynote’s chief technology officer. “You’ll find some websites coming up with homegrown algorithms to do this, such as ‘people who viewed this also viewed that’ or ‘people who bought this also bought that.’

“While they do offer some help, algorithms based on views and clicks alone can end up being pretty random. If it’s purchase-based, it’s more accurate, but very limiting. You want recommendations to tap into the entire shopping process and really help a person think through their purchase,” says Brave.

Shortly after implementing the program in fall 2007, Sun & Ski saw an immediate 59 percent lift in the average order on its website, and that figure held steady at 36 percent through the holiday shopping season, according to Scott Blair, Sun & Ski’s director of e-commerce.

An observer tag is deployed on all the web pages, which catalogs everything that users do on the site. All of the information is anonymous, with no IP address or specific demographic information collected. “We watch for key behaviors: what users search for, what links they click,” Brave says. “We don’t just watch whether they went to the pages and what they bought; that’s not enough. Did they scroll through the whole page? Did they add it to their cart? And we’re looking for connections between products and user interests. We do that for the thousands of users and we discover the patterns.

“When a new person comes to the site and starts exhibiting behaviors that we recognize, we can immediately connect them up with like-minded peers who have shared the same interest and recommend what they ultimately found of value,” he says.

More flexibility
Blair investigated a number of potential solutions, “but the big difference I saw with Baynote … was the ability to manage those recommendations to a certain degree.” For instance, if Sun & Ski adds a new product, “we can give it a bump for a little while to get it started. If we don’t like how a product recommendation is working, we can blacklist it. If someone’s looking at a jacket and we have a black pant that will work across the whole line, we may establish that as a recommendation for all the jackets.

“It really offered us a lot of flexibility inside the system that we can manage ourselves,” he says.

At Sun & Ski, the aim is to use the information to identify trends early in each of the distinct adventure seasons. “It’s hard to know what’s going to be the hot item for the season,” Blair says, “but if we can spot a trend that’s happening early, it may give opportunities to go back to the vendor and hit some home runs out of it.”