Boost attributes in recommendation results

Another interesting feature that allows you to influence the result of personalized recommendations is boosting the attributes that are used when adjusting the listing of products to the customer's expectations. Due to the higher importance of the selected product attributes, you can influence the order in which they are presented on the personalized recommendation listing.

Let's start with ready-to-use use cases that address the basic scenarios of personalized recommendations.

Promote or demote products depending on the margin

One of the overriding business goals is to maintain the highest possible level of margin on sales, which can sometimes conflict with recommendation algorithms. You can use margin information from your product feed to promote products with high margins and demote those with low margins.

Promote discounted items to customers at risk of churn

You can use the prediction module to create a broader customer context by predicting churn probability and making more advanced recommendations. Using this information, you can promote products at a discount to customers with a high likelihood of churn without limiting the scope of the offer.

Promote a brand in recommendations

Personalized recommendations can be an important part of product navigation. They allow building customer journey paths and can also be used as a paid e-merchandising space, where you can boost a specific brand by influencing its display position in personalized recommendations.

Promote customer's favorite brand

Use the knowledge about the customer's favorite brand to prepare personalized recommendations. Increase the relevance of the brand attribute by influencing the algorithm that suggests recommended products.

Inspirations

Raising or lowering the ranking of a product in recommendations is the least invasive way of adjusting personalized recommendations to sales scenarios developed by marketers. While maintaining the same number of products, we can strengthen or weaken their visibility in recommendations for a specific client.

Demote best-selling products

Depending on the adopted sales strategy, we may want to make the best, most attractive products to attract customers. E.g., through organic search (SEO), but at the same time, they were not promoted in a special way to maintain the linking as long as possible. Reverse boosting of such products will limit their presentation to the most "suitable" customers.

Boosting recently added items

Attaching customers to their favorite products may build a risk of losing them if we limit too much access to a well-known product by promoting other ones. However, it is worth persuading them successively to new products by gently increasing their score in recommendations for those who have according to the AI engine the greatest compliance of preferences with promoted novelties.

Boosting a new delivery of products that have previously sold out

Customers who were looking for a product, but due to its unavailability, were directed to the card of the product unavailable at the next visit and after replenishment of the stock may receive support (scenario) promoting the products they missed. Filtering personalized recommendations risks incorrectly defining the attribute that serves this restriction. If we are not sure if we define it well, or if our goal is only to slightly promote certain product features, we can use the boosting option, i.e., raising or lowering the position in the ranking due to the value of the attributes.

Boosting of products of the same brand as in the basket

The complementarity of products is very often associated with the brand. We often buy or build sets based on the collections of the same manufacturer. To communicate more effectively with customers behaving according to such a pattern, we can strengthen their personalized recommendations by using this attribute.

Boosting of the last items on stock (with the correct/matching size)

Create a recommendation based on the combination of size attributes with boosting products selling out, usually at a discounted price. Thanks to this combination, we can reach the right customer with the product without building additional personalized promotions.

Boosting of recycled/vegan products.

We are not always able to clearly state what value a specific customer attribute should have, and boosting is the safest solution. Lactose-free, vegan or recycled products for customers who bought one but others in the cart did not meet this condition is a very good scenario in this situation.

Dominik Królikowski, Business Value Services Director

For even more interesting Use Cases describing recommendations visit our Use Case Library.

For our Partners, we have prepared a series of workshops and inspirational meetings that expand knowledge about this functionality. If you are not a Synerise Partner yet, consider registering.