Problem
Unengaged injured workers didn't know how to proceed to complete their tasks after landing on the dashboard of our claim management app. Because these users couldn't complete their required items, clients wouldn't pay us for them.
Solution
I redesigned the Jarvis dashboard by centralizing the engagement experience and exposing required actions to reduce complexity and make it easier for injured workers to engage.
Impact
Discovered an engagement ceiling
After observing no change in engagement after a substantial overhaul to the dashboard, we found evidence of an in-app engagement ceiling, prompting us to shift our focus toward increasing our enrollment rate to improve engagement.
Improved Product Strategy
Through our enrollment research study, we uncovered a breakthrough approach: bypassing the enrollment barrier entirely by removing the need to log in. This strategy would allow users to engage with our product instantly, leading to a potential 6x increase in engaged users compared to our original approach.
Enhanced User Understanding
The enrollment research study illuminated deep bio-psycho-social factors behind injured workers' interactions with digital products. The "trust" framework that resulted from the study has served as a guide when creating new features to ensure user-centric design.
Our company's rapid growth was jeopardizing our bottom line. Clients paid us based on the number of users who completed engagement actions on our platform. With a 49% average engagement rate, acquiring more clients meant that as our volume of claims increased so would the loss of revenue from more unengaged users.
Insights from our product analytics platform showed unengaged users arrived to the dashboard upon sign-in but didn’t visit another page, meaning they never completed engagement actions (e.g., signing a document, sending a message, reading a resource, etc…) We hypothesized that injured workers didn't have enough clarity on what their responsibilities in the app and therefore were unable to complete the tasks.
Considering our goal and pain points, I chose three design principles to drive the iteration process:
Exposure: Injured workers need to have visibility into engagement actions and respective details when they enter to app for the first time.
Gamification: No one wants to deal with insurance; use familiar gamification techniques from analogous industries like banking to encourage interaction despite undesirability.
Centralization: Prevent injured workers from having to travel throughout the app by bringing all required actions to one place.
At the end of the iteration process, there were three versions of a new dashboard. The key distinction between the versions was the level of exposure; each version was an attempt to balance displaying required tasks to injured workers without overloading them with information, especially when they were using the app for the first time.
Dashboard versions
To measure each version’s efficacy, I conducted a remote “click test” with injured workers. Our most effective design had an 88% target hit rate for engagement actions, followed by a design with a 55% hit rate.
Although the "best" design was the one with the higher hit rate, conversations with the engineer and customer ops team exposed that the best design would not be able to handle the variety of engagement types and details between different client environments. Therefore we decided to move forward with our second-best design, which could account for these differences.
After six months of monitoring impact, we saw that the engagement rate of in-app users hadn't changed. We wondered how such a large redesign of the platform could result in no significant shift in our target metrics—positive or negative. The takeaway? We had reached a critical threshold where large efforts to change in-app engagement weren't producing worthwhile results.
The average engagement rate of 49% was a combination of activated user and unactivated user engagement (unactivated users could still engage by sending messages or completing questionnaires using our text service). This unactivated cohort was playing a huge role in our average as only 8% of users were engaging.
At the start of the initiative, the team decided to target activated users since previous efforts to improve unactivated engagement or converting users had proven difficult. However, we now had data showing that targeting in-app engagement was not fruitful. I proposed that if we could learn how to get people to sign up, then we could increase the volume of engaged users, and since activated users had an 89% engagement rate, greatly increasing of revenue.
I led a research study where I interviewed injured workers to learn about their experience, behaviors, and pain points around signing up for online healthcare/insurance platforms. Suffering from pain and distress, injured workers repeatedly expressed frustration with burdensome modern-day activation flows like repetitive information entry, recall of complex passwords, and laborious multi-factor authentication. They also expressed hesitancy with opening links from unrecognized senders and the need for an empathetic, end-to-end claim experience.
By grouping interview insights and using thematic analysis workshops, we developed a framework that illuminated how trust, and its three components–humanity of the experience, transparency of intent, and precision of information–drive injured worker enrollment. Using the trust framework, we audited Jarvis and revealed our largest barrier to injured worker enrollment: the login wall.
By requiring injured workers to activate an account to access our application, we had created a barrier between injured workers and the tasks they needed to complete. This not only increased the effort necessary for injured workers to engage in-app but also decreased trust as the login wall obscured the value of our application (transparency) and required repetitive information to be entered (precision), both resulting in a lower likelihood for injured workers to enroll and engage.
Existing activation to engagement flow
Typically, once a claim was opened, injured workers would receive an introductory email/text from Jarvis with a tokenized activation link allowing us to direct them to a client-specific activation page. By using tokenized links for task completion instead of activation, we would be able to remove the login wall and create a direct path between injured workers and the application’s value. In other words, click the link to engage, not sign up.
Proposed "Tokenized Engagement Link" flow
Although our initial approach didn't impact engagement as we had expected, the resulting pivot proved invaluable as it uncovered larger barriers to engagement, reshaped our product strategy, and enhanced our understanding of injured worker needs.
Copyright © 2024 Sebastian Viasus