Ecommerce giants like Amazon, eBay and Netflix have more than demonstrated the value of offering product recommendations. A wonderful tool for upselling and cross selling, recommendation engines are also perfect for tasks like reminding shoppers they need to make additional purchases to get the most from an item they just bought at your ecommerce store.

If you have yet to launch one on your site, you’re missing out.

Here’s how recommendation engines work.



Data Collection

Recommendation engines use both explicit and implicit data to develop their suggestions. Explicit data is an aggregation of likes, comments and other evidence derived from user inputs. Implicit data is acquired from customer actions such as order histories, return histories and the frequency at which items are placed into shopping carts. The number of pages views, clicks, and search events an item gets are considered as well.


Data Storage

Logic dictates the more traffic your site gets, the more accurate the results will become. These algorithms learn as they’re fed more information, so the more data you’re capable of storing, the better your recommendation engine will perform.

This makes the type of storage you employ critical to the successful operation of the functionality. Ideally, you’ll employ a scalable managed database. Whether you have electronics, cosmetics, furniture or ebooks online stores running on ecommerce platforms like those offered by Shopify, incorporating a fully-managed database capable of growing to accommodate you’re the amount of data you acquire will make your recommendation engine function more efficiently.


Data Analysis

The resulting recommendations can be broken into three categories, centered upon the method from which they are derived.

  1. Content-based recommendations are based upon the similarities a given product has to what the users’ behavior indicates they would prefer.
  2. Cluster recommendations (also referred to as Hybrid recommendations) are based upon the suitability of a product to accompany the chosen item, regardless of whether or not other users have considered it.
  3. Collaborative recommendations are made based upon preferences exhibited by other users—as gleaned from observing their “views” and “likes” of a given product.


As you may have guessed from the way they chose to title their offering, Amazon uses an item-to-item collaborative model based upon prior purchase behaviors.

“Customers who bought this item also bought…”

Amazon’s recommendation engine matches each of a user’s purchased and rated items to similar items, then combines each of the resulting items into a recommendation list for the user. The algorithm is capable of scaling to consider immense amounts of data, so it can provide extremely accurate recommendations in real time.


Why You Need One

Netflix estimates the value of its recommendation engine at one billion dollars. In a press release the company issued on the topic, a Netflix spokesperson writes:

“Consumer research suggests that a typical Netflix member loses interest after perhaps 60 to 90 seconds of choosing, having reviewed 10 to 20 titles (perhaps three in detail) on one or two screens. The user either finds something of interest or the risk of the user abandoning our service increases substantially.”

Long story short, Netflix knows it only has a minute and a half at most to get something in front of a viewer they’ll want to watch. With 80 percent of its viewer selections coming from the recommendation engine, the company estimates cancelled subscriptions would approach a billion dollars annually.


The Bottom Line?

If you haven’t added one to your ecommerce site already, it’s way past time to incorporate a recommendation engine. You’ll help your customers find products they want to buy more quickly. They’ll appreciate the personalized touch and your average order value will increase, thus expanding your revenues. It’s a scenario in which everyone wins.