It is no coincidence that Netflix matches your favorite films and series or that Amazon recommends products that match what you were looking for. The mechanism behind this intelligence system is called a recommendation engine. 

But what is this technology about? A recommendation engine is what is technically known as collaborative filtering. This type of filtering uses different algorithms to find matches between your previous searches and other products or services available on the website. 

This search analysis allows eCommerce to suggest items of similar interest to consumers until they find what fits their needs. Thanks to this technique, it is possible to achieve a perfect eCommerce-customer match.

As a Google Cloud Machine Learning specialist partner, we accompanied a healthcare eCommerce company in its digital transformation. This company needed to go further and relied on our team to apply machine learning technologies to their platform.

How to apply Machine Learning in eCommerce

We start by establishing a strategy for analyzing information about User Behavior on the website. This first step is key to logically applying this technology and achieving good results. 

The next step was to apply a data collection system to establish a score of interest for each product. Once we had collected all this information, we found out which items had caught the user's attention.

And now, how do we use this analysis in practice? Basically, by applying Machine Learning clustering techniques - an automatic grouping of products based on certain criteria - on GCP (Good Clinical Practice) products and then implementing a recommendation system based on collaborative filters on the website.

In short, by applying this mechanism, consumers will receive suggestions of similar products that have already been of interest to them previously. Therefore, we manage to recommend those products that are most likely to be relevant to the user.

Recommendation engine to improve conversion

The recommendation engine implementation on our client's website was carried out on both home and product pages. The KPIs improved significantly: the click-through rate provided by the engine was 25% higher than those achieved by the recommendation widgets previously installed on the website.

On the other hand, the depth of navigation increased by 12%. In other words, the number of clicks between pages increased, thus improving the relationship of eCommerce with the users’ interests. 

Similarly, there was a 10% decrease in the bounce rate on product landing pages. After the recommendation engine implementation, the browsing time increased: the users now decided to stay. And, last but not least, the overall conversion rate of the website improved by 0.5%. 

Ultimately, as these results show, we could improve the customer experience and achieve the goal of increasing the number of conversions.

Advantages of recommendation engines in eCommerce

The main potential of recommendation systems is that they eliminate pain points on both sides of the chain. On the one hand, eCommerce gets closer to the users' needs while they end up finding what they have in mind in a shorter time. 

This is its main advantage, but applying recommendation engines in eCommerce offers two other benefits related to sales.

Bounce rate reduction

Many eCommerce sites have a portfolio of hundreds and thousands of products. On the face of it, this should benefit the consumer as they will have more options to choose from. 

However, a large volume of items can sometimes be an impediment for the user to find what they want. In other words, it could become an obstacle to purchase. 

Machine Learning and recommendation engines are the key players in suggesting products to consumers based on their previous purchases and searches. With this:

• Decrease the search time spent by the user
• Minimize frustration with a failed search as much as possible
• It shows a selection of products that is useful to the customer

With this substantial improvement in the user experience, the foundations for customer loyalty and future prescription are laid. 

Product upselling

Recommendation engines perform a function similar to the shop staff who advise you in the store. Based on your tastes, your previous purchases and the contents of your shopping cart, they recommend other items that you may need.

This in-store advice service is one of the functions that eCommerce did not have until now. Now, this Machine Learning technique adds a touch of personalization and sales orientation to online businesses.

These recommendation systems facilitate upselling through different actions on the web -suggesting products that complement the item you have in the basket or playing with a price saving by adding that extra product to your final purchase-.

Recommendation engines are a good option for introducing Machine Learning in online commerce. From there, other AI solutions can be implemented, such as a virtual assistant.