The TV and web retailer will bring its e-commerce and video programming to France in 2015. QVC already sells in Germany, Italy and the ...
Sponsored Supplement April 2011 Getting personal
(Page 2 of 5)
Personalize by product
Understanding the attributes of the product being viewed is often just as critical to making product recommendations as understanding the shopper's preferences and previous shopping history. A consumer shopping for skis that are not sold with bindings will most likely need bindings. There is also a good chance she will need poles and boots.
"Consumers have preferences about the types of products they like, but products have their own attributes that influence what to recommend to the consumer," explains Ken Levy, CEO and co-founder of 4-Tell Inc., provider of personalized product recommendation services. "A lot of products are not sold as part of a package, but the customer needs to build a package that combines several items. Taking that attribute into consideration makes it possible to suggest the additional items needed to complete the package. When combined with what we do know about a shopper, it results in very powerful personalization."
With 4-Tell's system, a retailer can leverage what it knows about a particular brand, together with what it knows about a particular customer, to make relevant recommendations even if it has no previous history of selling a specific product. A retailer selling its first pair of K2 Apache skis, for instance, can look at what bindings, boots and poles are most frequently sold with K2 skis. In this way, 4-Tell's system more closely matches recommendations that would be provided by a salesperson in a bricks-and-mortar store.
"There is no one killer algorithm when it comes to personalization, rather it is the sum of all the parts—consumer preferences, product attributes and how consumers interact with products, brands and the retailer," says Levy. "We use every data point we can—including offline data if available."
Levy adds that 4-Tell makes relevant recommendations when only the product or the customer are known. "The best recommendations are made when both are known, but a recommendation based on one is still effective," says Levy. "The recommendation type is controlled by the retailer."
The growing sophistication of personalization engines is making it imperative for retailers to integrate them with their inventory management systems to avoid recommending products not in stock.
"Not all retailers want to recommend products that are out of stock," says Neil Hamilton, CEO of PredictiveIntent Ltd., provider of behavior merchandising and personalization solutions for e-commerce and m-commerce retailers. "Integrating personalization engines with inventory management makes it possible for retailers to recommend the next best product if the first choice is not in stock."
Integration with inventory management systems also makes it possible for retailers to recommend relevant items that carry higher margins, such as store-branded merchandise. "Store-branded items usually cost less, which is a benefit to the customer, and they deliver a higher margin to the retailer," says Hamilton. "If it is clear the consumer is brand-driven, our personalization engine can be instructed not to recommend store-branded products."
To determine how important brand is to a retailer's customers PredictiveIntent will run A/B tests recommending brand names and store-branded items, and see which performs best. "We even test to see which recommended items in a category have the highest conversion rate and factor that data back into our personalization algorithms," Hamilton says.
PredictiveIntent's Intent Prediction Server uses open application programming interfaces that allow retailers to integrate the personalization engine into all their other applications, regardless of their technology platform. Algorithms and filters can be customized to meet the retailer's business objectives.
Shopping is not the only reason consumers come to a retailer's web site, and it's important that retailers offer personalization to the customer conducting research. Many consumers visit a retailer's site to learn how a product works, find instructions for do-it-yourself projects or check out how to clean and care for an item before making a purchase. Recognizing this trend, many retailers have set up information sections within their web sites that include tutorial sections.
"Creating a bridge between the tutorial section and the web store so a consumer moving from the former to the latter lands on the product page for the item they were reading about is something retailers need to build into their personalization strategy," says Hamilton. "Personalization has to be integrated across the entire store."
Because research is becoming a bigger reason why consumers visit retailer sites, including product ratings with product recommendations, as well as links to customer reviews, can increase the relevancy of the recommendation.
"Retailers don't want to be thinking in terms of one or two product recommendation scenarios," says iGoDigital's Tobias. "Recommendation scenarios that include more information around the product can help improve conversions."
Retailers also want to adjust recommendations as consumers move through the purchase process. A consumer in the viewing stage can be shown a recommendation for products viewed by other consumers with similar preferences. When the shopper makes a purchase, the retailer can recommend products bought by consumers with similar preferences.
"The idea is to have a recommendation scenario geared to each action the consumer takes," adds Tobias. "Product recommendation can't be a one-size-fits-all approach."
The Customer Intelligence Engine from iGoDigital collects real-time information about a consumer's behavior patterns as she moves through a retailer's web site, creating a profile of her habits and preferences. The engine also takes into consideration how product attributes match up with a consumer's preferences as well as geographic, environmental, technological, seasonal, and other non-behavioral observations to determine product recommendations.
Once a consumer has purchased an item it is important to keep an open line of communication to her in order to generate future sales and build customer lifetime value. Personalized e-mails or mobile SMS text or MMS alerts enable retailers to reach consumers with relevant messages, including promotional information about products previously viewed or new arrivals or accessories that go with a recent purchase. These types of personalized, ongoing communications are better able to bring customers back to your site.