Hitmetrix - User behavior analytics & recording

Produce Spooktacular Results Through Personalization

Although done with good intention, personalization can sometimes make consumers uncomfortable. But when it offers shoppers genuine value, as opposed to being a prying eye, personalization can be more of a service and seem less like spooky witchcraft. For online costume and party supplies retailer BuySeasons, value proposition is working like magic as marketers leverage predictive analytics to deliver recommendations.

A hair-raising problem

BuySeasons has three brands: Birthday Express, which offers kids party supplies; Costume Express, which sells child and family friendly costumes; and BuyCostumes, which provides mostly adult costumes. While Birthday Express’s business is fairly consistent year-round, the models for Costume Express and BuyCostumes are significantly affected by seasonality. In fact Eric Kirkhofer, COO of BuySeasons, says that the number of Web visitors on each costume site typically skyrockets from the regular season’s 10,000 visitors per day to more than 200,000 visitors per day in the time leading up to Halloween.

“We’ll do probably 75% of our business in about a six-week period during mid-September,” he says.

Although the seasonal spike has its benefits, it also has challenges. Since the majority of shoppers visit the site only once a year, most aren’t familiar with the array of products the brand offers. So, most consumers don’t create an account or log into their account before shopping. And Kirkhofer says more than one third of visitors drop straight to the brands’ homepages to browse, as opposed to heading to a product details page to buy.

Lack of consistent, identifiable traffic makes it difficult for marketers at BuySeasons to segment their customers or generate personalized experiences.

“With a costume site that people don’t visit all year round, you don’t get enough traffic to be able to build a segmentation model,” Kirkhofer says, “because literally in a day in October, we’ll get as many visitors as we do in a month in March.”

Still, the company has given these efforts an honest try. Kirkhofer says that the company manually assorted costume skews. So, when a shopper clicked on a costume, the brands’ marketers would handpick three related costumes to show them. This, however, didn’t provide the contextualization that marketers needed.

“Just because I’m looking at a Batman costume, [that] doesn’t mean that I want to see a Superman costume,” Kirkhofer says. “It depends on what I looked at before that, my experience, and what I’ve done on the site.”

Kirkhofer wanted to find a way to drive personalization—one that would result in a minimum 10% increase in conversion without sacrificing average order value. “At the end of the day, personalization, the way we have approached it, is an improved merchandising initiative,” he notes. So, the company met with Reflektion in April and implemented its predictive analytics platform in July.

Recommendations that are dead on

The team at BuySeasons decided to test out the technology on BuyCostumes first. So, marketers started by implementing two Reflektion solutions: the Personalized Site Search tool and the Predictive Product Recommendations tool. 

First, there’s the site search tool. When a shopper starts typing keywords into BuyCostumes’ search bar, the brand will present that shopper with pictures of products she might be looking to buy. If the shopper is new to the site, the products will be based on the keywords and what’s trending. If the shopper has been engaging with the site, the brand will present options based on the items that she’s clicked on.

As for the Predictive Product Recommendations tool, BuyCostumes has a section on its site called “Just For You.” Again, if a shopper is new to the site, this section will feature costumes that are trending. However, if the shopper has been clicking around the site, then the platform will track her behavior through a cookie and serve up costumes that better reflect her interests. The company also has a “You May Also Like” section on its product details pages that features recommendations based on the same criteria.


All of the data collected goes straight to BuyCostumes’ e-commerce analytics platform.

Sweet results

With these new tools, Kirkhofer ran an A/B test for two weeks that provided the new experience to some consumers and the old experience to others. The COO says that the results were so good that he and his team decided to implement Reflektion’s technology across the other two brands immediately.

“We were well north of 20% on BuyCostumes in the month of July in terms of both revenue per visitor and a conversion rate lift,” he says.

And seemingly BuySeasons has been able to maintain its momentum throughout the Halloween season. After deploying another split test in September, BuySeasons found that it sustained its 20% lift in revenue per visitor and conversion across all three sites. Consumers also experienced 35% more pageviews when viewing the new experience versus the old experience.

A spirited future

So, what’s next for BuySeasons? Kirkhofer says that he’d like to expand the new experience beyond costumes and start personalizing the accessories shopping experience. As for lessons learned from his personalization pilgrimage, Kirkhofer says the takeaway is quite simple: Remember, the “path to product drives revenue.”

“You can’t buy if you’re not on a product details page,” he says. “And if it’s hard to get to a product details page and people give up the journey before they get there, they cannot buy.”

Related Posts