Variation sort order
Variations are different versions of the same base product, such as a T-shirt that comes in multiple colors, or a pair of shoes that comes in several sizes. So how does Constructor think about variations when it comes to product ranking?
The score of a base product is actually the score of its best variation. But what defines the “best” variation? Like all products, Constructor leverages real user behavioral signals such as clicks and purchases, as well as manual merchant controls such as boosts and buries, when assigning a ranking score.
You as the customer have many tools available to you to help influence the ranking of variations. Global searchandising rules, for example, apply to both the base item and each variation. Global rules encompass both boosts and buries that are set broadly for all search or browse pages, rather than on a query by query basis. What this means is that if there is a global boost for the color “blue”, and an item has a variation that it is blue, that base item will be boosted. If that item has many variations, one of which is blue, that item will be boosted with the blue variation being picked over other variations (such as pink).
Variations will subsequently sequence according to their respective scores. So in the example preceding, the blue variation may show as the face-out default, but then the remaining variations will also be ordered according to their assigned scores as well (which are determined using those same behavioral signals and manual merchant controls).
You can also leverage the following strategies to influence the default ordering of variations as well:
- Assign a ranking score by passing a
suggested_score
in thevariations.csv
- Identify a default variation by using
metadata:is_default
in thevariations.csv
If all ranking factors are the same, we prefer the order you've specified using suggested_score
and/or metadata:is_default
. Otherwise, we prefer the one with the highest score (e.g if only some variations are affected by searchandising rules or the affinity engine)
Additionally, we have an option to always respect the customers preferred variation. Under this option, we can also always respect the customer’s preference, as specified by suggested_score
& metadata:is_default
& ignore the ranking factors when picking the most relevant variation. Please reach out to your Customer Success Manager if you would like to enable this feature.
For additional information, please visit our REST API documentation here.
Updated about 2 months ago