Personalized Recommendations Drive Double-Digit Conversion Lift

By Matt Pillar, chief editor

July 2016 Innovative Retail Technologies

One-to-one customer engagement enabled by machine learning yields a 13 percent reduction in bounce rate and a 33 percent average order value improvement at Marmot.

A recent study from Gallup shed light on just how valuable it is to invest in customer engagement. Fully engaged customers, defined as those who have had a measurable reaction, connection, or experience with your brand, represent a 23 percent share of wallet, profitability, revenue, and relationship growth premium compared to average shoppers. By contrast, the study found that disengaged customers — those who have no emotional connection to your brand — represent a 13 percent discount in those same measures.

It’s no surprise that customer engagement and loyalty applications have been hot in retail, particularly e-commerce, where customer data is easier to gather and easier to analyze than in any other channel. Forrester recently reported that 78 percent of digital marketing experts were using collaborative filtering techniques (leveraging the known preferences of a group of users to make recommendations on the unknown preferences of others) and real-time self-learning analytics (the collection and analysis of contextual data — such as browsing history) to deliver targeted digital offers.

Let’s stop right there and be clear. While retail academics and analysts debate the true meaning of personalization, we can agree that its definition is not synonymous with the “wisdom of the crowd.” It matters not whether that wisdom was gleaned from a focus group or learned by a machine; it’s not, by definition, personal.

The wisdom-of-the-crowd approach is still the path to promotions and recommendations predominantly followed by most retailers. It’s the tack that was taken for years at Marmot, a brand that’s been designing high-end outdoor apparel and gear since 1974.

Much more recently, when Jeff Milbourn realized that Marmot’s direct-to-consumer e-commerce business was topping 5 percent of annual sales, the company set out to see if it could further improve the value of customer engagement by moving away from “many-to-one” promotions and toward a true “one-to-one” digital experience. Milbourn is director of e-commerce for Jarden Technical Apparel brands (Marmot, ExOffi cio, and Zoot Sports), a business unit of the former Jarden Corporation, and now part of Newell Rubbermaid. Jarden became Newell Brands in April of this year when the multi-billion dollar parent of the Marmot brand was acquired.

From Broad Segmentation To Personal Relevance
Milbourn says that until recently, product recommendations at Marmot were generated from a “mashing” together of customer personas. The approach resulted in specific offers being targeted to a segment of consumers. It bore fruit, but it wasn’t the brass ring of personalization that progressive retailers have been racing toward.

“The pilot revealed that placing the widgets on our key pages resulted in a 13 percent decrease in bounce rates and a 10 percent decrease in exit rates on those pages.”

Jeff Milbourn, director of e-commerce, Marmot

“We sought to create a more one-to-one experience by customizing personal interactions with the brand to make those interactions more relevant,” he says. “We felt that true personalization of the products we presented and the promotions we offered would promote an experience of discovery.” That discovery, he surmised, would increase engagement, and the revenue gains promised by research such as the aforementioned would follow.

In 2015, the e-commerce team at Marmot began its search for a personalized engagement solution. “We researched several personalization platforms, most of which were promising conversion lifts in the low single digits. We were looking for a lift greater than 10 percent,” says Milbourn. Late that year, Marmot and Milbourn became intrigued by the promise of Reflektion, a young company founded by former Google engineer Amar Chokhawala in 2012.

“Reflektion’s Individualized Commerce solution stood apart on two fronts,” explains Milbourn. “First, they guaranteed a 10 percent increase in conversion and demonstrated real-world performance that more than doubled that figure with existing clients. Second, we really liked their individualized Visual Site Search widget, which opens a personalized visual display as a shopper enters a search term or places their cursor over the search box.” The Reflektion platform leverages machine learning to understand individual consumer preferences as they click through an e-commerce site, then provides recommendations via the search widget based on those interests and actions.

Pilot Tests Exceed Expectations
With its vendor selected, Marmot set to work on implementing a pilot of Instant Visual Site Search and Predictive Product Merchandising, which presents customer-specific product recommendations on the merchant’s home page, provided by Reflektion. “This was a relatively painless process with very few steps to go live, which was another key to our selection,” says Milbourn. “Reflektion placed a couple of containers throughout our site to contain the site search widgets, and a JavaScript tracking pixel to gather traffic information that drives the merchandising solution.” Those code modifications made, Marmot connected a feed of its catalog to Reflektion, and the companies worked together to customize the look and feel of the widgets to Marmot brand specifications.

 Next, Reflektion conducted a scientific A/B test over a period of four weeks, during which time the vendor “dialed in” the individualization algorithm as traffic volume increased and it received feedback from the tracking pixel. At the beginning of the pilot, just 5 percent of site traffic was exposed to the Reflektion experience, a figure that climbed to 50 percent as the algorithm was adjusted. “Once our legacy site experience and Reflektion site experience hit a 50/50 mix, we were able to use Google Analytics to track, measure, and compare conversion rate, average order value, bounce and exit rates, and so on,” says Milbourn.

With the pilot in full swing, Marmot was given access to a Reflektion dashboard that reported on the performance of the platform in real time. Marmot could use that data to make adjustments to the placement of personalized search widgets on its site. As those adjustments were fine-tuned, Milbourn learned that placing the widgets on key pages was resulting in a 13 percent decrease in bounce rates and a 10 percent decrease in exit rates on those pages. “We also found that shoppers who put their cursor over the search widget produced 33 percent more revenue per visit than those who used the traditional search field and clicked on a resulting product.” When the test was concluded, Marmot had bested the 10 percent goal it set out to accomplish, with a 13 percent improvement in conversion — and a revenue-per-conversion lift of a full third compared to traditional search. Milbourn adds that payback was quickly and easily achieved; vendor remuneration is a simple contracted percentage of the retailer’s sales conversion gains.

In Search Of Constant Improvement
 Marmot has since rolled Reflektion Individualized Commerce Solutions out at scale, but the 13 percent conversion increase isn’t the end of the personalization story there. Because the Reflektion platform is constantly learning and responding to customer preferences for attributes like color, style, size, gender, and category in real time through click behavior, it drives continuous improvement by prioritizing the most individually relevant products and content first. That will continue to improve results at Marmot as the platform collects and refines its analysis of customer browsing data.

The platform is also revealing invaluable information to merchandisers and web designers at Marmot, which is helping them optimize storefronts, layouts, and promotions. At press time, Marmot was in the process of a site redesign, an effort guided in part by what it has learned from the Reflektion implementation. “Based on the results we’ve seen, we’ve learned a great deal about what drives higher conversions and average order values,” says Milbourn. “They’ve given us the tools to gain that insight, and now it’s our job to put what we’ve learned front and center via design and user experience improvements, like making the search bar more prominent and engaging throughout the site.”

Milbourn says the company will also turn its attention to the mobile experience and email personalization with the aid of Reflektion. “Everyone has played with personalized subject lines, but we send out a lot of product emails on a weekly basis,” he says. “The Reflektion platform serves up products that are relevant to specific recipients, not just personalized address fields, so we’ll be taking advantage of that.” On mobile devices, the platform delivers the same individualized widget tools available on large screens, tailored to fit specific devices. That’s an opportunity for differentiation in itself. Last year, a Pure Oxygen Labs report found that some 80 percent of the Internet Retailer Top 500 list had yet to migrate to responsive design for mobile commerce optimization. Mobile personalization puts early adopters light years ahead of that curve. Considering the growth of mobile commerce, which BI Intelligence predicts will account for 45 percent of digital commerce by 2020, investment here is a new retail business imperative.

Retailers that report they are highly effective at personalization make up an elite group. A recent report from Gartner revealed that less than 10 percent of tier-one retailers believe they are highly effective. Even more alarming, nearly one-third say they have limited or no capability to support personalization efforts. That spells a big opportunity for merchants like Marmot, who are differentiating through true one-to-one personalization.