Tips & Tricks: User Affinity Profiling
What is User Affinity
Every interaction a user has with your site or app provides signals about their interests, preferences, and intent. These signals can come in the form of an action, for example adding a product to the cart. They are also derived from the attributes of products. A user viewing products with the same attribute values indicates a preference to the color, style, price range, etc...
Dynamic Yield now enables you to target users based on this real time User Affinity Profile collected.
How is User Affinity Calculated
The Affinity profile is composed of attribute values and their scores. Every time a user interacts with a product, the score of each attribute value is updated by the type of interaction with it (adding the product to cart warrants a higher score than viewing it).
If a user adds a red dress to the cart, we will promote color ‘red’ and category ‘dress’. If she purchases the dress, additional weight will be given to those same attribute values (red and dress). Red and dress are both attribute values. Color and Category are the respective attributes (columns)*. Recency of the interaction also has an impact; the more recent the higher the impact, for example interactions that occur during the session, will have higher impact than ones that occurred a week ago.
Targeting Users Based on the Affinity Profile
Now, you can create audiences in order to target your customers based on their affinity profile. For example, a user with affinity to the color brown (value) and category women (attribute), would see a banner promoting brown boots on the homepage, and more brown items in a recommendation widget.
Setting up your Site for Affinity
In order to collect data and build User Affinity Profiles on your site, there are a few steps to take:
- Define which product attributes should be considered in the affinity profile calculation
- Specify which columns in your feed should be whitelisted to be used for Affinity calculation
- Work with your CSM or support in order to set this up
- Use columns in your feed with finite values/options
- Good examples: Categories, High level categories, Colors, Brands, Materials, Styles, price range, is_on_sale, is_new, etc...
- Bad examples: SKU’s, groupID’s, descriptions, names, price, etc..
- Create audiences based on top categories, strategic preferences, etc...
Inspiration for using Affinity on your site
- Replace the hero banner on the homepage
- For example, present a banner featuring brown boots for users with an affinity for the color brown
- Reorder the menu based on the user’s affinities
- For example, based on user-affinity, surface only the most relevant and personalized filters to each visitor, instead of the generic ones. Each user is granted quick-access to filtering by favorite-category, favorite-sport, favorite-color, etc...
- Reorder items on entire pages
- Serve targeted discount notifications to users with an affinity to certain products
*only product attributes (product feed columns) that are whitelisted will be scored for the User Affinity Profile
Click here to read more about Affinity Audiences
Please sign in to leave a comment.