Artificial Intelligence (AI) Is Here to Help You with Product Recommendations

By Vaclav Kamenicek in Development
·3 min read

Kentico EMS is a pretty extensible platform for executing diverse marketing scenarios. The scenario that resonates with organizations generating revenue through their digital e-commerce sites and applications is running the AI-powered product recommendation system to display products that customers are bound to like.

Connecting Kentico EMS with Recombee, a recommender as a service, enables marketers to incorporate product recommendation widgets tailored with an AI-powered engine throughout their sites or apps.

Generally speaking, Kentico EMS provides data and publishes recommended products from Recombee on a front end in the desired way. The way the recommendations are calculated is set and initialized through the Recombee web app.

What is it good for? We can pick these three reasons out of a solid bunch of other valuable ones:

  • Increase in conversion rates: the audience of users who clicked on a recommendation is more likely to buy something (eConsultancy).   
  • Increase in basket size: the average order value of visitors engaging with product recommendations tends to be higher (eConsultancy).
  • Positive impact on customer experience: no measures for that, but with personalized and relevant recommending in the right moments, you show that you care about your customers and that you know them.

Did you know? In 2014, 35% of products that consumers purchased on Amazon were already coming from product recommendations (McKinsey).

Recommending Capabilities and Models

Recombee is a real-time recommendation service, which means it returns the recommendations immediately after getting information about the user’s action on your website or in your app.  This is very important because you can deliver in-the-moment recommendations. For example, when your user adds a product to the shopping cart, you can recommend “frequently-bought-together products” on the next page.

Recombee provides you with recommendations based on behavioral patterns (so-called collaborative filtering) and on attributes of products and customers (so-called content-based recommendation). It appropriately combines the methods and adapts and tunes the recommendations based on the collected feedback when launched (there you go, artificial intelligence).

No Data, no State-of-the-Art Recommendations

Of course, you have to feed Recombee with proper data to make it work. Here comes Kentico EMS in to play. You can utilize all your user’s gathered data and tracked activities. For example, the system logs activities when visitors view pages, purchase products, abandon the shopping cart, register for events, or subscribe to newsletters.

Furthermore, you might want to feed Recombee with product information as well because it can use it for the content-based part of its recommendation models. Product status, category, variant, description, brand, or image (yes, Recombee has image processing), just to mention a few of them.

Kentico’s Recombee integration module gets this job done for you. It transfers selected data to Recombee, and, of course, you can extend the module to transmit a much broader flow of information if needed. You can download the module from Kentico’s MarketPlace for free and check out the code on GitHub

Reaching the Marketing KPIs with Right Placement, Design, and Copy

Once your data is all set up, how can you move from the starting point? A homepage might come first to your mind as it’s one of the most important places for product recommendations.  Other pretty powerful places for product recommendations are product categories, shopping carts, or out-of-stock pages.

Finding the right spot for recommendations on your site or in your app and setting the recommendation engine to provide outputs that make sense along the customer’s journey—those are important marketing jobs to be done if you want to fully leverage the power of product recommendations and reach the target KPIs.

Similarly, how you design and communicate product recommendations must be resonating with visitors to make them click. You might want to test which copy and design are generating the highest CTRs and the most conversions. Fortunately, there’s plenty of product recommendation tactics and best practices to be found throughout the Internet, which means you don’t have to start empty handed.

The integration module comes with an out-of-the-box product recommendation widget that can be easily manipulated by content editors and other non-technical users and placed throughout a site. Needless to say, you can further customize the widget or create a new one to fit the intended use and context.

dancing_goat.jpg

Ready for AI-driven Product Recommendation?

The integration module for Recombee is available at Kentico’s MarketPlace and GitHub for free, together with its description and installation instructions. Sign up for a free trial at Recombee.com and explore the possibilities of product recommendation powered by artificial intelligence within Kentico EMS.

Product recommendation is only one of the possible scenarios, and Recombee can be used in plenty of different ways with Kentico depending on the type of organization, context, and goals. Feel free to comment below and tell us which scenarios you are interested in or might be useful for your business.

By Vaclav Kamenicek in Development
Enjoyed this article? Check out even more development-focused posts at DevNet
Show Me
search
We're named a Challenger in the 2019
Gartner Magic Quadrant for WCM!
×