If you include a booster in your recipe, the visitor's affinity score is taken into account and items matching that affinity are boosted in the recommendations query results. The affinity score is based on how much a visitor interacts with specific items on your site (e.g. purchases, adds to cart, views items, views items for a length of time, and other interactions). For example, if your visitor shows a preference for a particular category of item, those that fall in that category will be shown first.
Booster Descriptions and Configurations
Each booster has several possible configurations which determines how each will affect the affinity score:
- Weight - affects how much the selected booster will be weighted in the recommendation query. The slider has five positions (1-5), each one being a factor by which the affinity score is multiplied
- Threshold - determines the point at which the visitor is considered affine for the booster. The slider has five positions (1-5), each one being a factor by which the affinity score is multiplied
- Lookback (days) - number of days to look back at visitor's history for this booster. For example, if a visitor bought a product three months ago, but the Lookback was set to 30 days, that purchase wouldn't be factored into the affinity score
- Item Types to Combine - item types that are to be boosted together (see description below for more information)
|Category||Categories of items or content the visitor shows interest in are scored higher|
|Department||Departments of items or content the visitor shows interest in are scored higher|
|Brands of items or content the visitor shows interest in are scored higher|
|Item classes the visitor shows interest in are scored higher|
|Style||Styles the visitor shows interest in are scored higher|
|Gender||Gender the visitor shows interest in is scored higher|
Combination boosters allow more specific items to be returned than a standard booster by item type. They work by looking at which items the visitor is affine to, and only boosting other items that share the same combination of categories and tags. Up to three separate item types can be combined together. If an item doesn’t have a value for that specific item type, it will not be boosted.
For example, with a standard recipe that has two boosters: a category booster and an style booster, a visitor that bought a red dress would be affine to both the style “red” and the category “dress”. This means boosted recommendation results are likely to be red, or dresses, but not necessarily red dresses. However, with a combination booster, only items with the specific combination of affine tags and categories are boosted. This means that red shoes and black dresses would not be boosted and would be significantly less likely to be returned in the recommendation results, while red dresses would be more likely.
This page has no comments.