The easiest way to make a sale is through a recommendation.
Unfortunately, many eCommerce store owners slap basic category level or best selling product recommendations on their site and never think about it again.
This is no longer the best we can do.
Today, product recommendation engines are capable of learning about a customer in real-time, and personalizing all pages they interact with - from home to checkout offers.
This post breaks down how product recommendation engines work, best merchandizing practices, and the results you can get through improving your eCommerce recommendations.
Types of Product Recommendation Engines
How do Product Recommendation Engines Work?
The purpose of product recommendations is twofold: first, to improve the shopping experience, and second to increase revenues.
Product recommendation systems do this by presenting shoppers offers they are most likely to want.
Engines sift through the tens, hundreds, or thousands of items a store carries, and decide which one best suits this particular user.
Generally speaking, there are three broad techniques engines use to filter through SKUs.
1. Collaborative Filtering Technique
Collaberative filtering uses other user’s actions to predict what another user will like.
For example, if one user brosed dresses, but ultimately bought a purse, the software would begin to draw a correlation between these two categories. As more and more users confirmed this association, it would begin influencing which products were being recommended.
2. Content Based Filtering Techniques
Content based filtering focuses on the specific shopper. The product recommendation software tracks a users actions, such as web pages viewed, products clicked on, time spent on various categories, and items added to cart.
Based on this information, a customer profile is created. This profile is then compared to the product catalogue to identify which items to show.
3. Hybrid Recommendations
The best recommendation software actually combines both techniques to give the most accurate prediction. This is how Barilliance works.
By combining both techniques, product recommendation engines are able to apply the "wisdom of the crowd" to prospects before they gather much data. As more information is learned about that particular user, recommendations become more and more personalized based on their session and use history.
Dynamic Product Recommendations: Don't use static product recommendations. Click Here to see how Barilliance personalizes recommendations on your home, category, and product pages.
In brick & mortar, stores are forced to choose a single merchandising strategy.
ECommerce stores don’t have this limitation.
Retailers can use personalization technology to create specific merchandizing strategies for any segment of customers. One of the main tools retailers use to accomplish this are product recommendations.
How Merchandizing and Product Recommendations Interact
By default, product recommendation engines work algorithmically.
However, the best engines allow retailers to "overrule" the software's recommendations in lue of explicit merchandizing rules you set up.
Retailers define what rules exist, and when these rules are triggered.
Again, using Barilliance as an example, you can determine customer segments that matter to your business. You can then selectively use merchandizing rules on these various segments.
As you can see, retailers have the capabilitity to define exactly what audience they want to display particular product recommendation widgets for.
By combining merchandising rules and product recommendations, you can create highly targeted offers.
You can promote best selling items to new site visitors, recently viewed items to returning visitors, and related products based on previous purchase to returning customers.
Product Recommendation Results & Stats
We performed an in-depth study of Barilliance customers who implemented our product recommendation solution.
The results were incredible.
Improving your product recommendations are the "low hanging fruit" of eCommerce personalization.
We host free demos for our product recommendation solution. If you would like to discover if we could improve your current recommendations with a hybrid, machine learning approach, click here.