
Improving Onboarding
Freddie Growth Team • Lead Product Designer • May - Aug 2023
Freddie patients now experience increased confidence in trusting us as their online healthcare provider. This series of successful growth experiments was launched over 3 months, resulting in a 22% increase in conversions across new user sign-ups, attended appointments, and reduced no-shows.
Overview
I instituted a series of high-impact experiments targeting the acquisition and activation stages in our onboarding flow, which resulted in a 22% increase in new patient acquisition across three months.
As the lead designer, I collaborated with Product, Engineering, Marketing, Customer Experience, and Clinicians to refine the onboarding process, which involved advocating for a balance of introducing new features and improving existing experiences.
Context
As an early-stage startup, we prioritized key features such as appointment booking and lab form uploads. Experiencing an exponential growth this quarter highlighted the need to improve onboarding. Our focus shifted towards optimizing onboarding processes, ensuring a smooth experience for patients creating accounts and booking consultations.
Goal
Boost user acquisition through executing growth experiments to increase sign-up rates and appointments attended.
Experiments
The following growth experiments were tested, iterated, and successfully shipped to 7,000+ patients.

Experiment 1
Questionnaire Descriptor Cards
The first onboarding step involves completing a risk screening questionnaire. This form requests personal and health information to establish a medical profile, determine eligibility, and assess financial assistance for PrEP.

Problem
From user interviews, we found that 20% of patients refrain from completing the sign-up process due to a lack of trust in Freddie as a virtual healthcare provider.

Goal
Strengthen patient trust during the onboarding questionnaire through transparent and intentional communication of data collection.
Decisions
Tone and Content
I incorporated a warm and welcoming tone inspired by the marketing content that resonated with patients to create a cohesive and familiar user experience.

Placement
I placed descriptor cards underneath each question as a form of reactive onboarding. This educates the 20% of users who have less trust with Freddie, but doesn’t force the majority to acknowledge it immediately.

Final Design
This experiment was launched as an A/B test. It outperformed the original, increasing questionnaire rates by 7.5% within two weeks.

Results also sparked a reassessment of the Freddie persona. Insights showed that patients are looking for more context in their medical journey, pointing to future growth experiments focused on personalization and information.
Experiment 2
Phone Number Verification
The second onboarding step involves booking an appointment with a clinician. The personal and health information given in the intake questionnaire is discussed and reviewed at this stage.

Problem
Amongst booking rates, data analysis revealed a noteworthy no-show rate of 25.9%. A significant portion was attributed to incorrect or inactive phone numbers.
Implementing a phone number verification flow became imperative to:
Capture a current missed opportunity for patients to engage with our clinicians
Save time and resources for clinical and operations teams
Elevate safety, security, and compliance measures
Considerations
Given the extensive interaction various teams have with patients, effective stakeholder management was the primary challenge.
Impact on downstream integrations: Numerous teams rely on external tools, such as Braze and Zendesk, for patient communication, where the existing phone number is stored. Phone numbers will only be added to or updated in these integrations after they are verified.
Approach to existing phone numbers: Existing phone numbers will be treated as verified numbers. Patients who have recent failed SMS's in Braze will be sent an email to prompt them to check and update their phone number in their profile.
Patient and provider-facing permissions: For V1, both the patient and Patient Care team can update a patient's phone number.
Goal
Implement a phone number verification flow for both patients and providers by executing a thorough testing, timeline, and release plan.
Decisions
Verification flow
I prioritized efficiency by utilizing industry-standard design flows to expedite wireframes into high fidelity prototypes.

This allowed me to focus on case-specific design details, such as setting the checkbox to a pre-selected state to minimize manual steps taken by the user, while ensuring that clinicians can initiate communication with patients.
Addressing use cases
I adapted the initial booking verification flow to suit the latter scenarios through higher-fidelity flows. This approach helped communicate updates to cross-functional teams.
Final Design
This project was implemented into the onboarding flow for 100% of Freddie patients, reducing no-show rates by 2%.

Experiment 3
Failed Log-In State
I improved the log-in experience for existing patients by redirecting them to their email without revealing/compromising private identifying information.

Problem
When a patient with an existing email attempts to log-in, they are met with the following message:

Freddie’s privacy policy prevents us from explicitly telling patients that they have an account. However, due to this lack of context, users have a low incentive to reach out to our support email because the issue is not clear.
Goal
Guide existing users to retrieve their account in a clear and discrete manner.
Decisions
Hierarchy
I identified an opportunity to establish a consistent error state, which was previously lacking. By integrating hierarchical elements like a header and icon, I aimed to bring more clarity to the intended outcome.
Content
I directed users to their email, since they would have received a notification while attempting to create a new account. This approach aligns with privacy policies, avoiding unintentional disclosure of a user's status as a potential patient.

Updating the design system
I incorporated a standard error message into our design system. The engineering effort was minimal since the update aligned with an ongoing experiment, providing an ideal opportunity to update our design system within a relevant environment.
Final Design
This experiment was launched as an A/B test. It outperformed the original page by 3% within one week.

Outcomes
All three experiments were tracked for one month post-release, resulting in:
7.5% increase in questionnaire conversion
3.3% increase in appointments booked per week
3% increase in total appointments completed
Final Takeaways
Growth experiments don’t have to be large in scope to be impactful. Vocalizing improvements to existing experiences and focusing on new experiences that directly address user pain points go a long way. Rapidly executing allowed me to iterate on learnings and build a roadmap for future growth experiments.