Objective
Vacasa would like to add a section to their abandoned cart email which recommends 3 units similar to the unit the customer abandoned in their cart. The objective is to create a programmatic solution to determining 3 similar units for the abandoned cart email, with the ultimate goal of maximizing conversion/revenue from the email.
Project Notebooks
Full Analysis Jupyter Notebook
Summary
I created a working KNN model used to determine similar rental units to a unit in a customer’s abandoned cart. However, to increase conversion, I can implement additional criteria for recommendations that may improve conversion. For example, I always want to guarantee that the units we’re recommending are in the same city or nearby to the unit in the customer’s abandoned cart. I would likely also want to ensure that my suggested units have at least an equal or higher maxoccupancy as the abandoned cart unit. Otherwise I run the risk of recommending a unit that doesn’t work for the customer. It might also help to have a similar baserate. And finally, equal values for dog, beachaccess, and hottub if the customer is looking for those things specifically. Once all of this has been implemented, I would run an AB test to determine whether the changes improve conversion significantly or not.
Leave a Comment