Many merchants use automated product recommendations to increase sales and conversions. These recommendations are usually dynamically generated on an ecommerce site, and they are typically based on the purchase habits of a particular customer, or a group of customers.
Strands Recommender is a leading provider of dynamically-generated product recommendations, and we recently asked its director of marketing, Trevor Legwinski, about the product-recommendation concept and its effectiveness to ecommerce merchants.
10 Questions on Product Recommendations
Practical eCommerce: What are product recommendations, and how do they work on a merchant’s site?
Trevor Legwinski: “For ecommerce companies with many products, structuring up-sells and cross-sells for optimal conversions can be very difficult.
“Product recommendations solve this problem by learning the behaviors of your site visitors on an individual and aggregate level. We automatically take this data and merchandise your site, matching behaviors and comparing intents of new visitors to thousands of shoppers who converted on your site. For returning visitors we display products based on their customer profiles, which are preferences our engine anonymously learns over time, including preferred categories, favorite brands, and even specific colors and sizes.”
PEC: Where do the recommendations typically appear on the site?
Legwinski: “Merchants typically place recommendations on their home pages, product detail pages and in their shopping carts. Additionally, we have started to introduce recommendations in the category pages, search pages and have even created custom “personalized for you landing pages” similar to Amazon.com.”
PEC: How do you determine which products to recommend?
Legwinski: “Assuming your company is a customer, our engine anonymously observes each customer that visits your site and builds a customer profile based on his or her shopping behaviors. These include, for example, products they visit, search, add to cart and ultimately purchase. For new visitors we display products that have had the highest chance of converting in the past for people exhibiting similar behaviors. For returning visitors we tap into their profiles and display products relevant to their preferences. Additionally, we provide easy-to-use merchandising tools that allow retailers to filter the recommendation results based on their own marketing and business needs.”
PEC: What sort of results have merchants seen that use product recommendations? Could you cite some actual data from your customers?
Legwinski: “We work with retailers in almost every category, from apparel and shoes, to electronics, books, games, music, furniture, and sporting equipment. On average they see an increase in sales in the range of 8 to 12 percent. Some of the more important variables affecting the recommender’s performance are traffic volume, number of recommendation widgets, placement of recommendations, and catalog size.
“To give you some specifics and put the numbers in perspective: an outdoor equipment customer that is using our $349 per month standard plan, has yearly revenues around $1 million and is currently recording an additional $2,000 a month in sales using our service. One of our larger apparel retailers utilizing our enterprise plan has annual revenues of $70 million and is making an additional $200,000 per month.”
PEC: Is the goal to increase the basic conversion rate, average order size or both?
Legwinski: “The goal of our recommender engine is to increase both average order size and conversion. By personalizing the experience to both new visitors and returning customers we are able to boost the number of products added to the shopping cart as well as the conversion levels of our customers. Additionally, using personalized emails to bring back existing customers will boost email response rates.”
PEC: Is your recommendations service hosted by your company? How does it integrate with a merchant’s site?
Legwinski: “Strands Recommender is a SaaS solution, which means it is fully hosted by us. This eliminates a lot of the version control and deployment complexities. A simple three-step process is used for integration:
- “Merchants upload their product catalogs into our system either via a file or a URL. This only has to be done once as we automatically refresh the catalog. The format is very similar to the same product feed you would send to a comparison-shopping engine. This feed helps our engine learn your product assortment and also stay up-to-date with your inventory. We also support data feeds from the leading rating and review providers, along with site search providers.
- “The second step is tracking user behavior on your site, which is then used to anonymously build customer profiles for each visitor. We can detect shopping patterns, correlate profiles with algorithms like “people that buy this also buy…” and other, more sophisticated, patterns. To do this, simple tracking codes similar to Google Analytics are pasted in the footer of your page templates in your home, product detail, shopping cart, order confirmation and, if applicable, “wish list” and “favorite” pages.
- “The last step is displaying recommendations to your visitors. Our customers select from a pre-configured set of recommendation widgets that can be quickly copied and pasted into any page of an ecommerce site. These widgets are based on best practices and generally work well for most of our customers. The system also offers a widget editor that allows customization of the look, feel and the contents of each widget. This way we ensure that each widget exactly matches your site and email templates.”
PEC: Does your service work with all shopping carts, hosted or licensed?
Legwinski: “Yes. Our JavaScript or API install makes it very easy for merchants with any shopping cart to install our software. We’ve taken simplicity a step further and have created shopping cart plug-ins for Miva Merchant and Magento stores.
“We recently opened up our API to developers, allowing them to create shopping cart plug-ins for their favorite carts and build interesting functionality on top of the user profiles and collective product behavior of the Strands Recommender system. We’re very excited to see the developer community embrace personalization and create some very useful and cool plug-ins.”
PEC: How much does it cost?
Legwinski: “Our plans start at $149 per month for smaller retailers and go up to $999, and more, per month for larger enterprise customers. Recommendation systems are becoming more and more affordable. In the past, this technology was only available to large retailers that could front the cost of thousands or tens of thousands of dollars per month. Our goal has been to make our technology affordable for smaller merchants.”
PEC: Does a merchant have to be a certain size, or have a certain number of SKUs, for recommendations to work well?
Legwinski: “We have found that retailers get the most benefit having at least 500 products and traffic of at least 15,000 unique visitors per month. This does not mean merchants with smaller catalogs cannot benefit from recommendations, but the real keys to success are traffic levels and product assortment. Recommender engines need a certain level of product data and traffic to be effective, and we’ve found 5,000 unique visitors is the lowest threshold in order to have a consistent, meaningful impact.”
PEC: What else do merchants need to know about product recommendations?
Legwinski: “One thing we didn’t emphasize much is the positive effect recommendations can have on email marketing. As a frequent online shopper myself I receive marketing emails from multiple retailers daily. But it frustrates me that many retailers are still sending a standard template that contains the same content for every customer. There is something to be said for retailers who take the time to personalize their emails for each customer and display a subset of products or promotions that they are going to care about. It shows the customer that you are really trying to service them, and, more often than not, they’ll buy more from you for putting in that extra effort.”