In April 2023, we doubled the conversion rate for our credit card application through a series of tweaks and one more significant launch.
- We changed the application’s background color from yellow (warning) to a more confident green.
- We removed the option to decline the card offer. The remaining option was “Accept” (or leave the page).
- Our initial application was iOS-based. We built and launched a web-based application in six weeks.
I was heavily involved with the tweaks. I was solely responsible for the web application from start to finish, including designing all screens, prioritizing work with Engineering, and coordinating a company-wide testing plan prior to release.
In May 2022, we launched the Ness Rewards app, a precursor to our credit card. It was taking us longer than anticipated to build the infrastructure for our card. Our goal was to launch the Rewards app and begin learning from our customers before launch.
With the Rewards app, customers could link their existing credit cards to our app through Plaid. We would then award them points for every dollar spent at health and wellness merchants. They could redeem those points in our curated marketplace of rewards.
This app offered little monetization (primarily through affiliate links). It did, however, afford us several advantages:
- Through the transaction data from linked cards, we could build up our Healthy Merchant Database, a proprietary list of health and wellness merchants.
- We could build our rewards marketplace and establish relationships with brands that would later become critical partners for our credit card.
- Analyzing the transaction data, we could validate our business’s core assumptions, such as the average customer’s monthly healthy spending.
In the fall of 2022, we launched the Ness Card. Since we had an iOS app available for customers, we built the initial application directly in the app. We assumed our initial cardholders would be current Rewards app users.
The credit card application became our primary funnel, and I reviewed the metrics daily using Mixpanel and Looker. Immediately, we noticed several problems resulting in a lower-than-expected conversion rate.
- There was a consistent dropoff in approved applicants after the hard pull. During a credit card application, customers typically go through a “soft pull” (no credit hit) and a subsequent “hard pull.” The latter can potentially impact your credit. Therefore, it’s logical to assume that customers proceeding through the hard pull are motivated to get the card. Following a hard pull inquiry, approved customers must accept the card offer, the final step. We noticed a significant dropoff between customers approved at the hard pull and those that accepted the offer.
- There’s a natural drop-off when you ask someone to download an app. This was a challenge for our Growth team. Across all of our marketing efforts, we had to ask customers to apply for a new (premium) credit card—no small feat! We also had to convince them to download our iOS app to access the application.
- Most customers only download a financial app after they establish a relationship. With established financial institutions (Chase, Wells Fargo, CapOne) and newer upstarts (X1, Bilt), customers are used to applying online and then downloading the app only if they’re accepted. In our case, we asked customers to download our app before getting a Ness Card. This creates a negative experience if the customer gets declined; they now have your app on their phone!
We improved our conversion rate and addressed the challenges above through smaller UI updates and a more significant launch.
In terms of UI, we made two meaningful updates:
- We switched the background color on the “Accept Offer” screen from a light yellow to a darker green. The assumption was that green signified calm and wealth, whereas yellow was closely associated with worry and caution.
- We removed the “No thanks” option from the “Accept Offer” page. Customers could either tap “Accept” or close the app entirely. Reviewing other leaders in the market, we discovered none of them had an opt-out option after the hard pull.
Regarding significant launches, we rebuilt our card application on the web to establish a customer relationship without first requiring an app download. We built the application in six weeks using React.
I’m specifically proud of several aspects:
- Using Firebase user authentication, users could start and resume their application at any point.
- We handled all application states, including frozen credit files, manual reviews, and SCRA freezes.
- We built a way for users to set their passwords directly during the application, drastically improving their onboarding experience.
- We launched 4 weeks ahead of schedule.
In combination, we more than doubled our conversion rate when examining the number of users that started an application to those that accepted the card offer.