Like a leaky bucket being carried across a room, every digital user journey has points where users encounter friction and leave—often right after you paid to acquire them through marketing or after you worked hard to attract them organically. The more holes you patch, the more users make it to where you are trying to take them. Opportunities for improvement are everywhere if you know where and how to look for them.

Over the past 5+ years my team has made dozens of site improvements—from where the user lands on our site, to how they find health insurance plans that meet their needs, and eventually to how they enroll in that specific plan. Complicating things further—we had to do this across both mobile and desktop experiences, because users' needs and behaviors are different across these device form-factors. We have been able to 3x site conversion by identifying and prioritizing the biggest opportunities.

Finding leaks

The more tools and data sources you use to uncover opportunities for improvement, the better. But leveraging them effectively takes hard work—and it’s never a one-person job.

The data needed is both qualitative and quantitative: User research, page level analytics, user session recordings, past a/b testing results, heat maps, competitive analysis, and site surveys. It helps to identify and focus on one prominent user type, documenting that one experience from end to end, overlaying any and all data and insights you have. If you are also able to do a full journey mapping exercise with these key personas, the biggest pain-points will surface quickly. To discover the smaller points of friction will take the collaborative effort of Product Designers, Product Managers, Content Designers, and Data Analysts.

Below I will highlight a few of the site improvements we made.

Top of the funnel

We designed and built out a platform for entry pages using a headless Wordpress instance that has allowed us to create a library of page modules and components that we are able to configure in numerous ways (think lego blocks) to develop pages for specific audiences and campaigns.

On these landing pages we have baked in dynamic elements that deliver a customized marketing message for each user's state and ZIP code, and echo the value proposition the user saw on the advertisement.

We are also able to dynamically toggle on/off telephone numbers, so that when we have spare capacity in our telephonic sales center we can drive more phone calls, and we are able to do this without hurting site conversion.

Plan shopping: Tools and layout

In order to make the best plan recommendation to our users we need some information from them about which doctors and drugs should be covered by a new plan. In our testing, users were always overly confident that if we just showed them all of the plans they could figure out which would be best for them, but the reality was that after 5 or 10 minutes users would be overwhelmed by the number of plans available and the complexity of distinguishing which would meet their needs best.

We knew that users that provided doctor and drug information converted at twice the rate of folks that didn’t, so we were constantly trying new ways to encourage people to use the tools that allowed them to supply this info. Again, the reality was that users that navigate to these tools are just higher intent and would likely convert at a higher level anyway (correlation vs. causation) and encouraging tool usage had diminishing returns.

When we moved the shopping filters from being vertical on the left-side of the page to horizontal across the top of the page above the plans, conversion went up 15% because they were more discoverable by the user.

Another big win came from encouraging plan comparison through the plan card design and page layout. We moved from a vertically stacked plan card to a smaller side-by-side plan card that allowed the user to view three plans right next to each other as they scrolled down the page and enabled them to flag plans that peaked their interest. Simplifying the card information through user testing and A/B testing enabled us to figure out what plan data was critical on this page and what could wait for the plan details page.

Enrollment

Once a user found the plan they thought was best for them they then had to enroll in that plan by filling out an application that is sent to the insurance carrier. State and federal laws required that we follow the exact question order and wording of a paper application the carrier submitted for approval, so we were very limited in what we could actually change.

That said—we were able to optimize the UX by breaking up the application into a few pages, making it feel more manageable to the user, introducing helper text to explain confusing insurance jargon, simplifying selections inputs, and leveraging auto-fill functionality that improved speed through the form and accuracy of the information.

Patched leaks

Through our testing and site improvements across the user journey we had several step-function increases in conversion, but a lot of results came from hard fought tweaks and tests that reduced a few clicks here or removed unnecessary friction there that enabled the user to more freely accomplish their goals. Throughout our testing we looked at page level metrics and overall conversion to be sure any improvements we made in one area weren't harming us somewhere else.

One additional thing I would mention is that we spent a lot of time working on the desktop and mobile experiences separately because what worked on one almost never worked on the other, so the two experiences diverged when needed. In the end—simplicity, removing unnecessary clicks and elements, removing distractions, and diligently following the data enabled us to find success.

One of our biggest challenges of being a health insurance marketplace and broker is customer retention. After we assist a client in finding and enrolling in a health insurance plan, they work directly with their healthcare provider and carrier. Over time—often a year or two—clients’ needs evolve, but by then, they may have forgotten who helped them find their plan in the first place.

Goal:

Increase customer retention by leveraging an AI-powered virtual agent to deliver personalized health insurance plan information, respond to coverage inquiries, and generate recommendations for cost savings and optimized plan selection.

Product Discovery

We utilized a tremendous amount of existing user research and insights to inform our initial app concept and performed multiple studies as we further refined the user experience and design. We did an extensive amount of competitive research to see what features and functionality existed in carrier and provider apps in the market.

There was a clear opportunity to meet a user need that was not being met by anyone else in the industry, that being — an independent resource that provides analysis and recommendations for health insurance plans across all carriers in a specific market as a user's healthcare needs change over time.

The average debt settlement client takes 3-4 years to finish their program, so having a place where they can get an update on what's happening is important to them. We also need a way to notify clients if an issue pops up or when a settlement approval is needed.

Problem:
Previous versions of the Client Dashboard didn't provide enough account information or program data, and created more questions than answers for our valued clients.

Goal:
• Provide clients program details online in real time, empowering them to manage their experience.
• Introduce alternative communication methods for clients to get their questions answered without having to call: Chat and messages.

Product Discovery

We started by pouring over analytical site data to see what users were doing on the existing Client Dashboard and were able to leverage hotjar heatmaps and user session videos to get even more insights. Next, we interviewed a number of clients that used the existing experience to see what they liked and didn't like, and suss out any unmet needs that could be addressed in the new design.

One thing we learned was that because the lifecycle of a client is so long (3-4 years), the dashboard played different roles depending on how long you have been in the program. We decided to create three different user segments based on time in the program to cater to their needs. Once we collected and analyzed enough user data we started developing concepts to validate and test.

Site surveys: 1
Remote interviews: 7
Unmoderated user testing: 25
Onsite user testing: 5
Employee focus groups: 2
Prototypes built: 1

Result

The final product has been a considerable success. The number of additional deposits has increased by 75%, which means clients are building funds more quickly, expediting their program.

We reduced customer service calls for clients that visited the dashboard by 24% and moved many other inquires to chat, which we introduced during the redesign. Between the reduction in call volume and efficiency of chat communications we were able to decrease the need for staffing in our call center.

The way we surfaced alerts in the new experience increased their visibility causing alert clearances to increase across the board: settlement authorization, insufficient funds and account number verification.

Note: Not shown in the mockups above are a slew of other features that added to the success of the Client Dashboard redesign, including: Draft modifications, enrolled debt details, FAQs, support content, message center, document vault, and more.

Launch: October 2018
Reduced call volume by 24%
Increased additional deposits by 75%

At Freedom Debt Relief, our main mechanism for acquiring prospective clients is through an online form where we collect information about the user and evaluate their debt situation. Our legacy experience only provided a basic program estimate with the hope that the user would call us or answer the phone when we called them. We had to nail the mobile experience because 76% of our traffic is smartphones.

Problem:
Our customer acquisition experience only provided a basic program estimate and did a poor job of educating the prospective client on the value of the debt settlement program which effected user engagement and intent.

Goal:
Increase the contact rate of prospective clients by providing a more robust program estimate with payment and savings, educating users on how debt settlement works, and making it easy for them to return later.

Product Discovery

Getting to know your user is key to building great products. We did this by recruiting people that looked like our users or were actually users on our website. Over a couple of months we built and iterated on two prototypes by putting each updated version in front of users to validate our approach.

Site surveys: 2
Remote interviews: 8
Unmoderated user testing: 20
Onsite user testing: 6
Employee focus groups: 2
Prototypes built: 2

Result

The final product was a success because it increased our two key metrics (contact rate and close rate) and was well received by users as it offers them a full estimate of their program and the information they need to make an educated decision. The program summary page does the heavy lifting, but the other pages explain how debt settlement works, what our clients are saying about us, and provides details about the debts that Freedom can help with.

Launch: June 2020
Increased contact rate by 12%
Increased close rate by 2%

Having a fully online personal loan origination experience that allows a prospective borrow to go through the entire loan origination process online without having to speak with a loan consultant has allowed the business to scale at a faster rate and improves unit economics.

Problem:
Our loan origination process was dependent on customers speaking with a loan consultant on the phone requiring more people to grow the business.

Goal:
Extend the Pre-qualification flow to include taking a consumer through the entire underwriting and origination process.

Product Discovery

The first thing we did was spend time listening to customer phone conversations with loan consultants to begin to understand user needs and motivations. We then did a series of user interviews of both non-clients and people that had gotten a loan from us, then continued the discovery process by building personas of our target users.

Remote interviews: 10
Unmoderated user testing: 14
Onsite user testing: 8
Prototypes built: 1

Loan Dashboard Screen

Mobile Web Screens

Results

Creating this full online origination experience has been a major milestone for the business and has allowed the business to cater to a segment of user that is comfortable and willing to go through the application and underwriting process online, without ever speaking to someone on the phone.

We continue to optimize this experience by analyzing user data and reducing friction in the user experience.

A client journey map isn't a product, but the creation of one was an important project that required hours of discovery and effort, and needs to be updated over time.

Problem: One asset that was missing when I first started at Freedom was a documented client journey from beginning to end. People could talk you through roughly what the experience was like and where users may experience highs and lows, but it was anecdotal and imprecise.

Solution: Map out the entire client journey of a debt settlement user.

Discovery

User surveys: 2
Remote client interviews: 12
Employee interviews: 10