Reduce drop rate of the onboarding flow

BAEMIN is a consumer food delivery platform operating in highly competitive, low-margin markets such as Vietnam and South Korea. Efficient user acquisition is critical for growth, where even small friction in onboarding can significantly impact conversion and customer acquisition cost.

At the time of this project, BAEMIN Vietnam was seeing an increasing drop-off rate (around 15%) during the first-time user onboarding flow, especially immediately after users opened the app for the first time.

This project aimed to reduce onboarding drop-off with a strong focus on:

• Quick execution and low engineering effort to maximize impact
• Marketing demands and broader business constraints
• Building user trust and clarity for first-time users

‍

Objectives

What we wanted to achieve:

• Improve usability and heuristic quality of the current onboarding UI
• Reduce drop-off at the first onboarding step, encouraging users to proceed into registration instead of abandoning at the landing screen
• Maintain the existing registration success conversion rate, ensuring improvements at the top of the funnel did not negatively impact downstream completion

Success Metrics

To evaluate the impact of the changes, we defined the following key metrics:

• Drop-off rate at the first onboarding step
  Target: reduce drop-off from ~10% to ~7%
• Registration completion conversion rate
  Target: maintain conversion at approximately 70%
• Customer Satisfaction (CSAT)
  Used as a qualitative signal to assess perceived clarity and ease of use after UI improvements

Together, these metrics balanced user experience improvements with business outcomes, ensuring gains in early engagement did not come at the cost of overall conversion.

Heuristic Review of the Existing Onboarding Flow

To get started, I conducted a heuristic review of the existing onboarding flow, focusing on the very first moments when users opened the app for the first time.

Data showed a significant portion of users dropping at the initial landing screen, so we examined this screen in detail to understand what might be causing early abandonment.

Key Issues Identified on the First Landing Screen

The review revealed that the current design did not clearly communicate value or guide users toward a meaningful next step:

• Unclear value proposition
  The headline “Exclusively on BAEMIN” did not clearly communicate what the product offers or why users should continue.
• Supporting message lacked visual priority
  The descriptive text “Enjoy many delicious dishes with incredible promotions only available on BAEMIN”explained the value, but its smaller size and lower visual hierarchy made it easy to overlook.
• Overwhelming progress indicator
  A 5-step progress stepper was shown upfront, which risked intimidating first-time users before they understood the benefit of completing onboarding.
• Weak and ambiguous CTA
  The primary CTA “Explore more” did not function as a true call to action. It lacked clarity about what users would see or gain by tapping it, reducing motivation to proceed.

‍

Together, these issues suggested that users were being asked to commit before understanding the value of the product, contributing to early drop-off.

In addition, visual accessibility issues further compounded the problem:

• Low contrast and poor readability
  The visual treatment of the registration steps used low-contrast text and subtle UI elements, making key information harder to read—especially in quick, first-time interactions. This increased cognitive load and friction at a moment where clarity and reassurance were critical.

‍

Data Analysis & Conversion Insights

To validate the heuristic findings, I analyzed available onboarding and authentication data to understand how users were performing across key entry points.

Conversion Rate: New Users

Overall conversion from app install to completed account setup was relatively low. Only 60%–70% of users successfully completed the onboarding and registration flow.

This confirmed that a significant portion of potential users were dropping before experiencing BAEMIN’s core value, placing their first order.

Returning Users: Password Recovery

The data also revealed a critical issue for returning users. In the forgot password recovery flow, approximately 25% of users failed to reset their password and subsequently abandoned the app.

This meant that even users with prior intent and familiarity were being blocked by friction in early access flows, leading to unnecessary churn and lost reactivation opportunities.

Together, these findings highlighted that onboarding challenges were not limited to first-time users, but extended to returning users attempting to re-enter the product.

‍

Competitive Analysis

To better contextualize BAEMIN’s onboarding challenges, I conducted a competitive analysis across both local and international food delivery apps, focusing specifically on how they handle first-time onboarding, permissions, and registration.

BAEMIN’s Existing Flow

BAEMIN’s onboarding experience was structured into two distinct parts: Onboarding and Registration.

• Onboarding phase:
  Users were asked for key system permissions such as Location and Notifications. Unlike many competitors, BAEMIN attempted to explain why these permissions were needed, rather than relying solely on system pop-ups.
• Registration phase:
  Users were required to complete a multi-step process including:
  • Phone number input
  • OTP verification
  • Name entry
  • Password setup
  • Email input
  • App tracking permission (iOS only)

Only after completing all steps would users land on the home screen.

While well-intentioned, this resulted in a long and effort-heavy first-time experience, especially before users had a chance to perceive value.

Local Competitors (Vietnam Market)

Now, Grab, Gojek

Most locally operating apps in Vietnam shared similar patterns:

• Permissions for location and notifications were requested very early
• Requests were often delivered via system pop-ups, with little or no contextual explanation
• The experience felt intrusive and transactional, prioritizing operational needs over user comfort

Despite this being a common market pattern, it also normalized early friction for users.

‍

‍

International Benchmarks

Deliveroo, Uber Eats

In contrast, apps operating outside of Vietnam demonstrated a noticeably different approach:

• A stronger emphasis on welcoming users upon first open
• Clear, friendly explanation of value before asking for permissions
• Location access was framed as a benefit, not a requirement
• Permission requests were delayed until users understood why they mattered

These products treated onboarding as a trust-building journey, rather than a checklist of system requirements.

‍

Key Takeaway

The analysis revealed an important mismatch between design intent and market reality.

BAEMIN’s onboarding design, flow, and permission-handling approach were actually closer to international best practices — patient, explanatory, and value-driven. However, BAEMIN was operating in the Vietnam market, where user behavior had already been shaped by dominant local competitors.

As a result, many users were likely less patient with extended explanations and multi-step permission requests during first launch. What worked well in international contexts risked becoming friction in a market accustomed to faster, more direct flows.

Assumption: To reduce early drop-off, BAEMIN may need to selectively adapt its onboarding approach to better align with local user expectations, even if that meant deviating from international patterns and following certain local practices.

This assumption directly informed subsequent design decisions around sequencing, messaging, and permission timing.

‍

Key Design Challenges

1. Low-margin constraint
   Any solution had to avoid increasing delivery, support, or engineering costs.
2. Operational dependencies
   Accurate address data is critical, but asking too early caused drop-off.
3. Trust building for first-time users
   Users needed reassurance before committing personal information.

Based primarily on the first and second hypotheses, we made a deliberate decision to rethink the sequence of the onboarding flow.

1. Reversing the Onboarding Order

Instead of asking for system permissions upfront, BAEMIN would allow users to register first, postponing permission requests until they became contextually relevant.

This shift aimed to:

• Reduce immediate friction at first app launch
• Avoid overwhelming users with permission dialogs before they perceived value
• Align better with local user expectations shaped by competing apps

2. Adopting Local Registration Patterns

To further reduce friction, the registration flow was redesigned to follow local market practices:

• Simple and fast interaction model
• Minimal required fields upfront
• Clear, single-purpose steps

The goal was not to copy competitors directly, but to meet users where they already were, lowering the cognitive and emotional cost of getting started.

3. Improving Visual Clarity & Accessibility

In parallel with flow changes, we addressed visual friction in the registration UI.

The existing design relied heavily on a mint-colored background paired with light, low-contrast text, which made form fields and instructions harder to read—especially during quick, one-handed mobile interactions.

Assumption: Low visual contrast increased cognitive load and hesitation during registration, contributing to drop-off.

Design changes focused on:

• Increasing text and input contrast for better readability
• Strengthening visual hierarchy to clearly indicate primary actions
• Making form progression more scannable and confidence-inspiring

These adjustments aimed to reduce unnecessary friction at a moment where users were already investing effort into registration.

I decided to change the brand mint background, even though it is the brand color, but in registration, we need the design to be clear and easy to navigate. Additionally, white and small text on mint background is low contrast and hard to read as well

Rollout Strategy

To closely monitor impact and minimize risk, the new onboarding and registration experience was rolled out gradually:

• iOS: 100% rollout over 1 week to validate early signals
• Android: staged rollout at 10% → 30% → 50% → 100%, increasing week by week

This approach allowed the team to track behavioral changes, identify anomalies early, and build confidence before full deployment.

Key Results (First Phase)

The first rollout phase showed clear improvements across all primary success metrics:

• Registration completion rate
  Increased from ~70% to ~78%
• Drop-off after first screen
  Reduced from ~10% to ~5%, exceeding the original target
• Password recovery drop-off
  Reduced from ~25% to ~10%, significantly improving reactivation for returning users

Business Impact

Beyond funnel metrics, the improved onboarding experience also drove meaningful business outcomes:

• +20% increase in new user sign-ups compared to the previous version
• Improved re-engagement from returning users who had previously failed password recovery

These results validated the core hypothesis that reducing early friction—both in flow sequencing and visual clarity—could materially improve conversion without harming downstream performance.