TFT : A TOEFL Speaking Prep App


12 weeks | Published in App Store in December 2017

Team and role

TFT (Task Forced Team) is a startup based at San Fransisco, who provides TOEFL prep classes facing the Chinese mainland market. TFT developed a free website to collect and share English speaking materials and planned to develop a mobile app when I joined the team.I work on the team as the mobile product designer, responsible for the App UX and UI design




The task is to design a handy and go-to App for TOEFL candidates to practice speaking. The final deliverable of this stage is a Minimum Viable Product (MVP). TOEFL is a test that all non-native English speakers have to take before applying for universities in the US.

Problem zone

The existed TFT web is not convenient nor adaptable for many use cases.   


We plan that the function of the App is limited to onboarding, practicing and account management.

Case study

VoiceTube: learn about the active study and user engagement.

Duolingo: the language study on mobile.

SoundCloud: the featured audio player.

User research

I synthesize user information from CEO, the customer service record, user observation, and interviews. We bring in users in the head start since we already get the goodness of having hundreds of students as leading users.

I create target personas under two main categories: the TOEFL candidate, for Accessible Design, and the English learner for Inclusive Design.


I use card sorting to arrange information architecture.

I draw journal maps, mind maps and wireframes to decide the shortest user flow.

I generate high-fidelity mocks for each design review using Sketch and Zeplin.

Design review

I hosted meetings on a weekly basis. Over the meeting, I first present my design or solution. Then CEO, developers, and the web designer would provide me with comments and feedbacks. Brainstorming and problem solving closely follow up. After the meeting, I will justify and validate the solutions by research or test.

User interview

We build a chat group to conduct 1 on 2 interviews to further validate my design. I provide more than one solution for hypothetical features and conduct qualitative tests to gather decision supports. We build user suggestion list to keep valuable but unprioritized ideas.

Build - measure - learn

I refine our prototype after each meeting and research to keep the design process agile.

I document all feedbacks and solutions for better collaboration.

Engineers participate in every design review so we work in a nonlinear, efficient and lean design-development process.

Determine the future strategic direction

After brainstorming, we broaden our vision, picturing a language materials library using multiple media, and aiming to create a platform where users can immerse and learn English by deliberate practice. To help general learners learn English are our essential product ethos.



Guest Users

Browse all topics and mock questions

Browse and listen to other users’ answers

Search speaking questions with keywords.

Filter questions with task number, TPO number, and category.

Browse our course info

Login Users

Could post text answers plus audio answers

Could check profile page of his own or others

Could listen to sample answers under selected topics, offered by our native-speaker couch

Could enroll in courses or register for seminars

Email notification for course or seminar enrolled

Go-Premium Users

Could take courses.

Browse daily write-ups

Submit homework

Check feedback of their homework in the profile page

Certified Coach

Could group users and monitor their updates.

Could review their answers



I had direct access to observing our target users online. The capacity of TOEFL prep classes offered by TFT is normally around 400 people. We have a large WeChat group and also small chat groups of 10 where students submit 60s audio answers.


I brainstormed with CEO at the beginning to come up with creative features.

For instance, unlike normal horizontal audio play bar, I came up with a featured idea that the audio bar could be displayed horizontally to visualize comments and couch reviews.


01 Candidates usually have limited materials to talk about during the exam.

02 Mobility provides more convenience and inclusion for not only the TOEFL candidates but also the larger group of English learners. It provides another option to kill free time.

03 Test candidates wish to listen to “full mark” answers and implant high-quality answers. The assignment group of 10 wouldn’t meet their needs.

04 Answers posted on WeChat groups are hard to retrieve or keep.

05 The website is functional for TFT classes, but a large amount of audio answers posted under each topic are unorganized and overwhelming for all users.

06 Being rooted on WeChat brings goodness like easy acquisition and message function, so the App design should consider WeChat into part of user flow.


At this stage, I turn all the requirements, research findings and brainstorming ideas, which by this moment are all hypotheses, into design. I used card sorting to decide the Information Architecture, and journal map to decide the user flow.

I hosted the team meetings frequently to review and revise my design and further discuss the viability of the hypotheses.

Take the iteration of the tap bar as an example. In my first draft, I designed “Speaking”, “Saved”, “Notifications” and “Profile” taps. After a couple of meetings, only three of them are left.

User test

The method I used is visualizing the uncertainties through small AB tests and threw the illustration of tests in the group chat for feedbacks.

For instance, I tested if the list view of all answers under a question is necessary; if users prefer to browse or record audio answers on their mobile devices and etc.


What can be improved?

01 Be more inclusive:

be the go-to platform for General English Learners, not only test candidates, which means we should bring in users in more various English levels. My strategy is to design for edge use cases and then extend to many.

For example, we have added open Chinese transcription for each topic but it would be closed in the next sprint, to turn accessibility into inclusion.

02 Be scalable:

the English language test prep market is huge. The DUA of TFT App is around 3k a month after launching without any marketing. Problems that we will run into include the user-generated content quality management, the community operation and so on.

Microsoft Research Asia ships an English study app in 2016 on WeChat platform using voice AI technology. In comparison, the large scale user base product is more data-driven, and more intelligent with all the big head goodness.