Algorithmic Sort

  • Problem

    Product ordering on the PLPs are manually ordered by Merchandisers then automated by date. Typically only the top 5% of PLPs are merchandised meaning the remaining 95% of PLPs are missing merchandising opportunities.

  • Requirements

    Create a scalable means of merchandising or sorting the remaining 95% of PLPs with merchandising-conscious sorting.

  • Solution

    Enable Merchandisers to apply merchandising preferences in a scalable and customisable manner.

Figma concept mockup of tool Algorithmic Sort tool (working with NDA)

Designing Algorithmic Sort

As well as supporting Merchandiser productivity, this tool would influence the majority of our customers’ experience of listing pages. We needed to ensure that is supports both our internal users as well as our end-users.

The two most important considerations when designing this tool is to deliver scalable ways of working, and to build a tool that supports business priorities without losing sight of the customer.

Scalability

Rather than design from the ground up, we matched how Merchandisers already think about ranking and sorting - the logic they use is already one-to-many, so providing opportunities to recycle algorithms was essential.

Keeping Customer Focus

The choice of metrics that influence the sorting was intentionally limited. This is not to say there wasn’t a lot of metrics in the weightings, however, not allowing Merchandisers access to all weightings was a decision made to ensure that all decisions benefit customers: a product’s margin has no influence on the customer experience.