Background

What is this?

Ladder’s mission is to protect families by helping people get covered through a policy. Still, we've had setbacks matching everyone with the best policy, which is not great for both our users and Ladder. In 2023, our Machine Learning team developed a model to improve these matches. This project showcases how I designed a solution that utilizes this model.

My role

I led the design of this project from May 2023 to July 2023.

In addition, I worked alongside a UX Content Designer, 2 Engineers, and a Product Manager.

My tasks entailed owning the end-to-end design process, defining product strategy, and pairing with developers.

Problems to solve

Ladder has long grappled with the persistent issue of attempting to cover users who don't buy a policy (i.e. convert), leading to high operational costs. Although users opt out of purchasing policies for various reasons, the primary deterrent is the pricing. Essentially, Ladder offers a policy to a user and the user thinks the price is too high and ends up leaving. This creates problems for both our users and Ladder:

  1. User problem: For users, they end up taking the time to go through our flow and decide to leave without a policy and leave their families unprotected. This affects our mission.
  2. Business problem: For Ladder, we end up incurring operational costs to try to cover a user who ends up not converting. This affects our profitability.

Machine learning model

Historically, we've always strived to address issues related to conversion and profitability. However, the predictive model developed by our ML team introduced a fresh opportunity for us to devise an entirely new solution.

This predictive model looks at various inputs to make an educated guess about whether a user is likely to buy a policy. Through backtesting we found that a certain cohort of users who decide not to purchase policies are actually costing us money when we consider all the different expenses.