problem

Covenant Eyes is a complex product. Since it works to protect its users from inappropriate content online, it has to embed itself pretty deeply into its users systems to gain the required levels of access and sovereignty needed to deliver a comprehensive solution. As a result, proper configuration of both the software and the user’s account is critical from the outset. For almost 20 years, the solution to this problem was a very lengthy signup process that combined account signup, billing setup, and software configuration. The process was very long (about 10 pages), confusing, prone to user errors, and performed poorly with regard to conversions. In 2019, as a result of partnering with an outside marketing firm, the company decided to overhaul the signup process to rectify some of these issues.

role

I was the product manager for the signup process during this time, leading a designer, 1-2 developers, and a QA person to accomplish this piece of work.

approach

I started this project by consulting with stakeholders to learn the business goals and constraints for a new signup process and their level of priority. I also factored in data from analytics on the current signup process to learn where the current dropouts in the signup funnel were and to learn the most common issues that user’s had getting through the process with an account correctly constructed and with proper configurations for their software installation.

Early on, I made the decision to separate the concerns in this process so that we could properly focus on solving the two main tasks we were trying to accomplish in signup - signup and onboarding. Signup was now going to only handle account creation/billing and then hand users off to a new onboarding funnel which would be handled by signup completion feedback, email, and our membership portal.

results

We built out a rough solution quickly and ran the new process as an A/B test (using VWO). We also utilized the heatmap features of VWO to gain greater insight into how users were interacting with the new process. We ran a variety of different types of user tests. Also, I established an Account Health performance metric to compare the ‘correctness’ of accounts created under each system. I ran rapid iteration cycles for about 6 months, learning along the way and dialing in the new process to meet the performance criteria we had set. After that time, we moved all traffic to the new process. Overall, the new process brought on an 8% increase in conversions and exceeded performance with regard to account health metrics in all categories.