Reducing Task Selection Time and Churn Rate in New Users
Time
H3 2022 - H1 2022
My role
Senior Product Designer
Team
Scope
Product design, visual design, product strategy, user research
Background
Toloka is a crowdsourced data labelling platform enabling fast & cost-efficient ML/AI training for companies like Amazon, Alibaba and more
As a senior product designer, I was responsible for the Web and Mobile applications for data labellers
How it started
The initial task assigned to me was redesigning the legacy user interface (UI) of the Data Labellers app, but I had hypothesized that the problem was much deeper than that
After an early conversation with the manager, I found out that the team thought that by changing the UI, they could reduce the high user churn rate
My first step was an in-depth analysis of the product, including key metrics and current issues, to determine how design could most effectively enhance the user experience
I also spoke with the support team, reviewed user complaints on social media, and examined feedback in app stores
Here are the issues I identified
The conversion rate of new users from registration to completed tasks is very low
Users experience difficulty in selecting tasks, often spending 3 to 5 minutes without success
The task interface lacks clarity, particularly regarding the required number of tasks needed to progress from the training stage to the exam and then to paid tasks
Users are uncertain why there are no paid tasks available after completing the training and exam, which leads them to avoid tasks with preliminary stages
The app's performance is not optimized, especially with a weak internet connection, which affects task availability
The interface design is outdated and has been criticized by both the product team and users
What do we, as a business, want?
To reduce expenses on user acquisition while expanding our core audience
What is the task?
To improve conversion rates by turning new users into those who successfully complete tasks and earn
How will we measure the result?
Reduction in the churn rate of new users
Increase in the average number of tasks completed by a user
Growth in the average earnings of a data annotator per session
Reduction in the time taken by a user to select a task
After discussing with the team, we decided to proceed with small steps, continually checking ourselves by looking at business and user metrics
Hypothesis 1
Transparency in the path from free training to paid tasks will increase conversion and speed up users' earning start
How will we test it?
After discussions with managers and developers, we decided to select a group of tasks in which we will limit the number of training and exam tasks
Results?
There was a significant increase in the number of users reaching paid tasks and in user earnings.
The speed of data annotation for clients increased.
What we learned
Users are willing to complete training tasks when they are informed about their number and the potential earnings they can achieve afterwards.
Hypothesis 2
Personalized recommendations can help reduce the time it takes to select tasks
How will we test it?
After discussions with managers and developers, we decided to select a group of tasks in which we will limit the number of training and exam tasks
Result
The time it took to select tasks was reduced by a third, and users actively began to use this feature
The number of tasks completed by an individual user increased
With these results, we concluded that we had sufficient compelling evidence to start redesigning the application
I created a new architecture for the app, incorporating all the insights we gained from the experiments and the users' objectives
Let me introduce the New Toloka App
How will we test it?
There was a significant increase in the number of users reaching paid tasks and in user earnings.
The speed of data annotation for clients increased.
Explore more
Increasing order completion rates
2021
Yandex Services
Car Wash reservation
2023
Yandex Services
🔒 Case study soon