Affinity

Dynamic Yield’s affinity-based personalization is an efficient way of untangling and leveraging the complex set of data surrounding user preferences. It allows you to deliver compelling product recommendations and highly relevant promotions.

Every user interaction provides signals about a user’s interests, preferences, and intent. For example, if a user added a product to their cart, we can learn that they are interested in some or all of the attributes of the product that was added to cart (the category, the color, the brand, etc.).

Combining product attributes, engagement types (view, add to cart, purchase) and their recency, allows Dynamic Yield to build an affinity profile in real time that maps how interested each user is in each attribute.

How Does It Work?

Step 1: Collection

For each engagement type, Dynamic Yield collects all the product attribute values that the user interacted with. In the context of a single visitor, we can imagine the following table:

The table above describes the following session: A visitor generated 4 pageviews in which the product color was green, 6 in which the product was red and 2 that were blue. All viewed items were T shirts for men priced under $50. He added two items to cart (red and green T shirts) but ultimately bought a single red T shirt.

Step 2: Calculating

The score sums up the total engagements for each value, and weighted by engagement type and engagement recency.

Engagement Type Weight

Each interaction type (e.g., product view, add to wishlist, etc.) receives a coefficient determined by an assumed level of intent of that interaction. For example, making a purchase receives more weight than adding something to the cart, which receives more weight than product views. When calculating the attribute value score, the occurrence of each engagement type is multiplied by the engagement type coefficient.

∑ Engagement type weight x attribute value count = score

Let’s assume the following coefficients: purchase weight is 4, add to cart weight is 2 and product view weight is 1. The total score for attribute value “color:red” is (1x6)+(2x1)+(4x1) = 12

Engagement Recency Weight

To identify user intent in real time as well as preference changes over time, each interaction receives a recency coefficient in order to favor recent activity over historical activity. There are three supported recency periods:

• Real-time: the last 48 hours including the current browsing session
• Recent history: the last 30 days
• Old history: the last 6 months

When calculating an attribute value score, score is multiplied by recency weight too.

∑ recency weight (engagement type weight x attribute value count) = score

Let’s assume the same user, from our previous example, comes back to the site 2 weeks later and purchases the green T-shirt he abandoned in the cart last time. Assuming that the recency coefficients are 8 for real time interaction and 2 for recent interactions the calculation of a total score for color green is now: 2(1x4)+2(2x1)+2(4x0) + 8(4x1)= 44

How Can You Leverage User Affinity?

The affinity profile of a user can be used on several ways:

• Recommendations: a recommendation algorithm, that allows you to recommend products according to each user's past behavior. For example, if the user purchased a lot of black shirts - the algorithm will recommend her more black shirts.
• Behavioral Targeting: a targeting and audience condition that allows you to target users who have a strong affinity to a specific value of a specific attribute (e.g. high affinity to black color).