Rocket Mortgage

Predictive Consumer Algorithms

Senior Product Owner, Data Intelligence•2020 – 2023

My debut project with Rocket was standing up a new data science team, designed to maximize the conversion of mixed media spend. This team worked on social media and display spend optimization algorithms, pushing us to compete with Google and improve overall spend-to-conversion rates.

The task:

Combine our internal datasets with social and display ad spend + conversion performance to create our own optimization algorithms
Implement the solution in an automated way to control spend as granularly as possible

The approach:

Brainstorm with data scientists on what existing datasets could strengthen the model, and what types of regression and modeling were best suited for the use case
Automate Google data ingestion, along with model execution
Automate updates to Google and social spend based on model outputs
Continuously refine and iterate

The challenge:

For me, part of the challenge was building a brand-new team inside a company I had just joined. I coordinated with other data science teams to understand the strengths of individuals across the department, and over time my team lead and I assembled the right group to take this on.

Learnings:

This was my first deep foray into data science, and it gave me real insight into what it takes to build a model from scratch
Building a new team is hard when you're new to the department
Data scientists are some of the sharpest people I've worked with — each one was incredibly talented
Operationalizing the process into hourly execution within our media spend pipeline proved to be as much of a challenge as building the model itself