Twyn
Twyn is a completely new way to practise speaking English. The app uses voice recognition and AI to create English conversations, using videos of real people. The product aims to allow users to practise ‘Speaking’ regularly rather than just reading words or tapping buttons on a screen.
Role
Competitor Research, Research Analysis, UX/UI Design, Usability Testing, Stakeholder management.
IMPACT
Increase the number of conversation completion by 22%
Date
2023
Background
The opportunity to be a part of an early start-up company in the EdTech field has been a fantastic challenge for me. In my role as the first designer on the app following the MVP launch, it was essential to implement processes and best practices to obtain user insights. This included handling user data safely, research planning documents and consent forms.
Problem
At present, users enter the app through a conversational English speaker. In addition to being intimidating and off-putting, English speakers with good language knowledge can also freeze at this point if they don't have a grasp on the language. In order to empower users when they speak to the Twyn in English, how might we help them feel empowered.
Solution
Give users the opportunity to find the correct answer so they can speak to the Twyn and move along in the conversation.
How we got there
I set up a range of data points to get the most accurate insight into user behaviour. This was achieved through surveys, analysing data in Mixpanel, setting up user interviews and ethnography sessions to observe users' experience of the app. With this data we established a sense of who our users were and what their goals and pain points were when speaking English. I identified the following pain points:
Feeling nervous when speaking English, even in the app
Not feeling confident about the answers they gave
Feeling like there was no improvement in their speech when using the app regularly.
Interestingly, users who accessed the app had high level reading and listening skills but lacked the ability to formulate their answers in speech. We needed to help our users when they could not move ahead with the conversations challenge.
Using the depth of data from regular interviews with users and conducting market research I set up workshops with senior stakeholders to create Twyn’s main user group personas. Creating a clear sharable artifact helped the team stay on track with solving core problems for this user group.
Personas
Persona developed from user research data to help us align to their core needs
“I understand what is being said but it’s too hard to answer the question”
Twyn user
I facilitated crazy 8 sessions with the wider team to explore design solutions that could translate into high fidelity prototypes to begin testing on users. This was broken into testing and iterating on iconography, wording and UI. Working with the PM we decided to test the design solutions on the highest performing conversation and measure its performance.
Key concepts & Testing
Sketches and research results
Our goal, as a company, is to encourage users to speak English as much as they can. Designing the 'Hint' feature allows the user to focus on their speech rather than forming a sentence. Here the user can click on a ‘hint’ button. This will show the answers they can provide when it's their turn to speak.
Conversation flow with hints
Results
The key metric that we wanted to improve was time in conversation and completion rate.
+22%
conversation Completion
-44%
page closure
Positive
User satisfaction
Features Roadmap
With our user insights and data points we have been able to identify a range of features that would help to make Twyn a one of a kind English Speaking app, this includes:
Improved response using Speech recognition
Simplifying task breakdown
Allowing users to see the task during a conversation
Simplify trouble shooting with microphone