Driving User Growth
Optimizing the Onboarding Workflow at TradeUp
Team
2 Data Scientists, 2 Frontend Developers
Timeline
August ‘23 - November ‘23
Task: This past year, I worked as a data science intern for eight months at TradeUp, a startup primarily focused on trading cards. During my tenure, my responsibility was to explore new ways to acquire more users, which led me to focus on improving the sign-up process.
Problem: With TradeUp relying on sporadic advertising, the site experienced a surge to 4,250 users, only to see a decline to 2,250 users the following month when there was no ad spend. This fluctuation highlights the importance of shifting focus towards sustainable strategies for user growth.
Analytics + Proposed Plan: Through analyzing user engagement data across different workflows, I identified several issues with the current sign-up process, notably malfunctioning components, indicating the necessity for a new approach. I proposed and implemented a progressive profiling strategy, simplifying the onboarding process by initially requesting minimal information and gradually collecting more data as users engaged with the platform.
New Metrics: Between September and October, the TradeUp marketplace witnessed a remarkable completion rate of sign-ups, nearly tripling from 27% to 87%. Moreover, October welcomed nearly 160 new users to the site, a stark increase compared to the 32 users in September.
By implementing this new method, I was able to directly addressed the non-advertised growth challenge, ensuring sustained user acquisition and retention.
What is TradeUp?
Established in 2020, TradeUp emerged from frustrations with the limitations and risks of traditional online trading card platforms.
Originating from encounters with restrictions on prominent marketplaces like eBay and encountering scams on platforms such as Facebook and Instagram, the founders discerned a pressing necessity for a more secure and structured trading environment.
By introducing innovative features like profile setup, card uploading, and communication tools, TradeUp fosters seamless interactions among collectors. These features ensure that users can build trustworthy profiles, easily share and verify trading cards with multiple parties, and communicate directly within the platform, reducing the risks associated with traditional online trading.
X’s Profile Page: TradeUp's user interface enables users to create engaging profiles, showcasing their trading cards and interests.
01
TradeUp is a secure online trading card platform that emerged from the founders' frustrations with traditional marketplaces like eBay and the scams on Facebook and Instagram. As a data science intern, I analyzed user sessions to address the instability caused by sporadic advertising and to develop sustainable strategies for consistent user growth and retention.
Overview
TradeUp vs. Other Trading Platforms
My Task
As a data science intern, I focused on growing the number of users on the TradeUp’s marketplace, which is one of of TradeUp’s biggest challenges. Previously, TradeUp depended on sporadic advertising to expand its user base.
However this led to significant fluctuations in traffic, with unique sessions — a user’s sequence of actions on TradeUp’s website – surging by around 4250 sessions during the month where the company would run temporary ads, only to fall down to normal numbers of 2250 users in the subsequent month where ad spend decreased to $0.
This pattern indicates that the costly ad investments was yielding minimal long-term impact, as user numbers reverted to normal levels in the following month.
More importantly, it also underscores the importance of shifting focus towards sustainable strategies for user growth.
To address this instability and find a more sustainable growth strategy, I began analyzing user sessions as a core metric. By understanding the patterns and behaviors of users during these sessions, I aimed to identify opportunities to improve user retention and engagement, thereby driving a more consistent increase in the number of active users on the platform.
When tasked with understanding and improving traffic fluctuations, one critical aspect to consider is the user journey from initial interaction to conversion. The sign-up workflow, or onboarding process, is a pivotal stage where users transition from being visitors to active participants on the platform. By examining this process first, we gained insights into potential bottlenecks or friction points that may cause user drop-offs.
To understand the sign-up workflow and the possible issues, I first delved into looking at the user interaction between all the pages. The sign-up process is separated into four pages: username, user categorization, value categorization, and trading preferences setup.
02
Processing Missing Data
To address traffic fluctuations and improve user retention, I examined the sign-up workflow on TradeUp, using Bubble.io and MixPanel to index components and identify bottlenecks, thereby identifying potential issues and collecting valuable user demographic data.
Username
User Categorization
Value Categorization
Trading Preferences
My primary tool of identifying these interactions relied on MixPanel.
However, due to minimal indexed components on the pages, pinpointing where users were dropping off and what they were clicking on proved challenging.Therefore, the initial step I took was to index all these components using Bubble.io, the platform utilized for building the website and transmitting all the data to MixPanel.
This indexing would serve a dual purpose:
Identifying potential technical issues that could hinder the user experience
Collecting detailed user data to understand the trading demographics prevalent within the TradeUp marketplace
As the data began to accumulate, I delved into MixPanel to search for correlations or clearer drop-offs in specific workflow components. The report was targeted towards the trading demographic: hobbyists seeking trade values ranging from $1 to $50 and filtered to a 30-day span.
This approach made sense as the goal of fixing the onboarding workflow was also to attract potential new users who might be less experienced with trading platforms, as opposed to those who frequent established online marketplaces.
Generalized Workflow
03
MixPanel Analytics
As data accumulated, I used MixPanel to identify drop-offs in the sign-up workflow, revealing a significant 74.3% drop-off from the Value Categorization to the Trading Preferences Page. This was occurring due to technical glitches and unresponsive buttons that were restricting users from finishing the sign-up process.
I began with creating a generalized workflow analysis, focusing on input-to-output metrics to assess the number of users completing the sign-up process.
Page-to-Page Workflow
To address this, I developed a more detailed workflow analysis, tracking user flow on a page-by-page basis to pinpoint specific drop-off points within the sign-up process. This refined analysis provided deeper insights into the exact stages where users were disengaging, enabling me to pinpoint precisely where drop-offs occur in the sign-up process.
The page-to-page workflow spans across three pages from the sign-up process, specifically the User Categorization, Value Categorization, and Trading Preferences pages.
This initial analysis of the generalized workflow revealed an alarmingly low user retention rate of 12%, highlighting significant user attrition before they could engage with the marketplace.
Important Takeaways from Page-to-Page Workflow:
The User Categorization Page was working correctly, as evidenced by the minimal user drop-off of one specifically bring in the metrics
There was a steep 74.3% drop off from the Value Categorization Page the to the Trading Preferences Page. Out of the 39 users who landed on the Value Categorization Page from the User Categorization page, only 10 users made it to the Trading Preferences Page.
The Trading Preferences Page appeared to have other distinct workflows, which seemed illogical given that this page primarily involved written input selection.
These patterns from the workflows suggested to me the presence of a technical glitch within the workflow, possibly redirecting users to incorrect pages or malfunctioning button components that hindered progression.
Hidden Discrepancies
Following discussions with our Bubble development team, we identified several issues affecting the user onboarding process. There were numerous reports of unresponsive (“stagnant”) buttons and various workflow discrepancies that allowed a subset of users to bypass certain onboarding steps without providing necessary responses. This flaw in the onboarding journey not only confused users but also led to inconsistencies in user data.
Specifically, it resulted in some users not being properly identified within the Bubble platform’s user dataset, contributing to discrepancies in the analytics reported by MixPanel. The feedback underscored the need for immediate corrective measures to ensure a smoother onboarding process and more accurate data tracking.
I proposed making the sign-up process optional to enhance the onboarding process and address user drop-offs effectively. The primary idea behind this proposal was to only require an email address. By doing so, users would have the flexibility to skip certain steps and directly proceed to becoming a member, thereby potentially reducing drop-offs between the workflows.
04
Aligned Proposal
To address user drop-offs, I proposed making the sign-up process optional, requiring only an email address initially, to streamline onboarding. While this reduced workflow friction, it raised concerns about losing essential user information, leading us to adopt a progressive profiling approach and simplify the onboarding process from four to two pages, balancing user convenience with data collection.
Initial Proposed Plan
While this approached massively simplified the onboarding experience, there would be too much essential user information lost that are vital for facilitating marketplace engagement.
Concerns about Data Loss
Why does this work?
This approach ensures that critical user data is captured while minimizing friction in the onboarding process. It strikes a balance between user convenience and data collection, ultimately enhancing the overall user experience on the platform.
Upon reviewing the data, it became apparent that the highest drop-off rates were occurring on the Value Categorization Page and Trading Preferences Page, which primarily focused on collecting user data. Analysis suggested that these steps might be perceived as tedious or could potentially deter users from completing the onboarding process.
Final Proposed Plan
This process involves initially requesting minimal information, such as name and username, to allow users to start using the platform.
This phased approach allows users to become gradually more invested in the platform while minimizing the cognitive load and potential for overwhelm.
Simplified to a Two-Page Onboarding Process
To address the concerns regarding omitting pages, we proposed transforming them into smaller sections within the user’s profile. Here, users would have the option to add features or provide additional information at their discretion. By integrating these changes, we effectively reduced the workflow from four pages to two, focusing solely on gathering necessary information — namely, the user’s name and desired username.
The omitted pages can be found under a user’s profile as smaller sections
Progressive Profiling Approach
After implementing the revised onboarding process, I revisited MixPanel to evaluate user engagement in October. The data revealed a notable improvement in our retention rate within the onboarding process. Out of 105 users who initiated the process, 91 successfully completed it, resulting in an impressive 87% sign-up success rate. This marked enhancement in user conversion is indicative of a significant reduction in churn rate.
05
KPI Metrics Post-Implementation
Post-revised sign-up process implementation, there was a significant positive impact on user sign-ups, with an impressive 87% sign-up success rate compared to the original 27%, and nearly 157 new users compared to the previous month's 32 users.
Updated Sign-Up Statistics
3. User Sign-ups
Calculated through total new user sign-ups.
September Sign-ups: 32 Accounts Created
Comparing the month prior to the implementation of the new onboarding process (September) with the following month (October), the results had become clearer. In September, we witnessed the creation of 32 new accounts. However, with the streamlined onboarding process introduced in October, the number of new members surged to 157. This exponential growth in user acquisition underscores the effectiveness of our strategic overhaul.
September vs. October User Sign-Ups
September (Left) vs. October (Right)
KPI Metrics
TradeUp has three KPI that they focus on:
1.Total Marketplace value (GMV)
How many listings and value are added every day? GMV is calculated through the total listing value.
September GMV: $12, 710.92
2. Transaction Volume
How many trades are being transacted ? Transaction volume is calculated through the total number of trading offers — active or ended.
September Transaction Volume: 335 Offers
In this project, my focus on optimizing the sign-up workflow directly correlated with the user sign-up metric, as it aimed to enhance the user experience and reduce drop-offs during the sign-up process. By implementing a progressive profiling approach and refining the onboarding process, we observed a significant increase in sign-up success, reflected in the surge of new members joining TradeUp. This improvement underscores the pivotal role of user sign-ups in driving platform growth and engagement.