Open navigation

Filtering Recommendations Using Product Attributes

This feature is available as part of the Smart Recommendation subscription.

Filtering Products in Klevu recommendations will allow you to skew the Klevu AI & strategy to better target products to customers. Using product attributes, you can now filter out products using attribute facets.

How does filtering by attribute work?

Consider a scenario where you are setting up Similar Products for the product page and you’d like to add complex filtering on the recommended products.

Use Case:

  1. If a shopper visits a product page whose brand is Adidas, then you’d like to recommend other similar products only from the brand Adidas that also have a rating greater than 4.
  2. If a shopper visits a product page whose brand is Nike, then you’d like to recommend other similar products only from the brand Nike that have Free Shipping but wish to exclude any products that have ratings less than 3.
  3. If the same shopper visits any other product page (other than the brand Adidas OR Nike), then you’d like to recommend other similar products from the same brand as the shopper is currently viewing & their price is in the range of +/- 10% of the product currently being viewed.

The above use case can be summarized as below:

Using Filter Rules & Filtering by Attribute, the above use case can be achieved in Klevu Merchant Center.

Filter by Attribute

Once a recommendation has been created, you can set up a rule or edit the default rule to apply filter by attribute. Based on your requirement, choose which attribute you’d like to use for filtering (e.g. brand, colour etc.).

Start by clicking on NEW FILTER:

1. Filter by Attribute on Homepage or Category Page

For a given rule, add a new filter. You can choose to either INCLUDE or EXCLUDE products using product attributes. Choose the attribute and its values to apply the filtering. Once set up, click on DONE to save and see a preview of the filters in real time. You can add up to 5 filters by attribute actions.

2. Filter by Attribute on Product Page

For a given rule on a product page recommendation, add a new filter. You can choose to either INCLUDE or EXCLUDE recommended products using product attributes.

You can set up the filtering actions as described above for the category or home page. However, for product page recommendations, you can also use a relative context. A relative context allows Klevu recommendation to read the attributes of the product that is currently being viewed by the shopper and dynamically pass that attribute to the recommendation for filtering.

For e.g. when a shopper lands on a shoe product page with the colour attribute yellow, Klevu can recommend other yellow shoes. But if the shopper changes the colour variant of the shoe to white, Klevu will take white into account and recommend other similar shoes that have the colour white. This is explained visually below:

To set up relative context filtering, after selecting the filter by attribute, select the “Current Product Value” in from the operator dropdown as shown below.

You can also set up a generic filter based on the relative percentage of the current product’s base price. For example, if you set up a rule for a relative price constraint of 30%, then if a product’s base price is $100, the filtered products would be limited to the range of $70 to $130.

You can configure filtering actions on the product page, as explained earlier, by utilizing a relative context. Extending this feature further, you now have the ability to read attribute values from the product viewed by the shopper, dynamically pass those attribute values, and match them with other attributes to filter the recommendations.

For example, when a shopper visits a mobile product page, you may want to display compatible mobile covers for the viewed mobile. However, if the attribute name differs between the two categories, such as "Model" for Mobile and "Compatible Model" for Phone Case, you can still recommend products based on the context of the viewing product using the Match With Value option. This allows you to provide relevant recommendations even when the attribute names differ between the products you intend to display.

Following is the visual journey:

To configure relative context filtering with the Match With Value option, follow these steps after selecting the attribute filter: choose Match With Value from the operator dropdown, as shown below.

Once you have chosen the Match With Value option, you should proceed by selecting the attribute that you wish to match the value with in the filtered products. In this case, the value of the "Model" attribute needs to be matched with the "Compatible Model" attribute.

Once you click on DONE, you will see the products you filtered in the preview section. Once you are happy with the rules setup, click on the SAVE & PUBLISH button (top right) to publish the changes to your storefront. Please note that it may take up to 10 minutes for changes to be reflected on your storefront.

See also: Pining or excluding specific products in smart recommendation.


Did you find it helpful? Yes No

Send feedback
Sorry we couldn't be helpful. Help us improve this article with your feedback.