Learning English on auto-pilot

After a year of research, I decided to go full-time on building @LockcardApp with @xinyifu in September 2021.

Lockcard is a unique IOS app that leverages your constant need to check your phone, to teach you stuff via notifications. Currently, it's being used to learn English vocabulary. Our goal is to plug into more intimate parts of your day and create a holistic learning experience. 

[hiring founding members, including a technical cofounder - DM me]


June 2020 - September 2020

Problem Context

Xinyi had been learning French for a while, but her initial enthusiasm was dying out as she became increasingly preoccupied with other activities. She found it increasingly difficult to take in new language data, especially memorizing vocabulary.

Project Hypothesis

We believe the journey to learning a language is long and arduous, making users feel like they are progressing very slowly after a while (aka the learning plateau). We are also aware from secondary research that young learners have an increasingly lower attention span, making it very hard to recall information. 

Project Goal

The goal is to understand vocabulary learners more holistically, their motivations and fears in life, then to investigate their experience with existing apps so that we can remove those painpoints and make the learning experience more meaningful and effective. 

User Interviews

We conducted 12 user interviews with people who had reached an intermediate level in their secondary language. You can view all our user research on Figma.

Interview Insights

From the interviews, we were able to dress three personas that would help us empathize with our future end-users beyond their experience on the app: life-long learners, casual learners and exam-oriented learner. They are distinct in motivation, cognitive activities (e.g. planning, reviewing, solving problems and measuring progress) and behaviours. You can view all personas on Figma.

Focused Persona: Casual Learner

Some of the key learnings for the focused personas are:

  • They have been learning the language for a long time, and have reached an advanced level.
  • They are busy with a job or high-level studies.
  • They are not focused on actively improving their vocabulary, but are used to searching in a dictionary when in doubt.
  • They are addicted to their smartphones.

Here's their journey map

Competitive Analysis

We have learnt a lot about our competitors' products, especially Duolingo which has an unusually high user retention rate (20%) compared to the average education app (3%).

Apps like Duolingo tend to position themselves as one-stop-shops to learning languages, for absolute beginners; we call these 'proactive'​ apps. Their teaching relies on gamification to keep users' attention high. Using a mix of pop-up notifications, rewards and social features, these apps play on our laziness and sometimes feel like they could replace the need to go outside and expose ourselves to organic learning in uncomfortable environments, like abroad or among a group of foreign friends.

Our approach to this language learning problem is more specialized and therefore complementary to existing products [Read our full story here].



Provide context

Many apps, especially Standard Language tests (e.g. IELTS/TOEFL), push a lot of information we feel disconnected from​ ​— nouns we don't need, adjectives we'll never know when to use — and in that process, they lose us because we don't create as meaningful an association as when a friend uses a word we don't know in a private conversation.

Offer a workspace

When our target users encounter new words, they lack a method for organizing them (printed notes vs note taking app vs dictionary). They end up losing track of what they have and haven’t reviewed, creating only more frustration. View our UserJourney Map in the Appendix.

Help Memorize

As a combination of the above two pain points, we lack the energy or the time needed to review new vocabulary consistently, resulting in a never-ending cycle of Googling the same words every time.


September 2020 - Ongoing

The very first sketches of the app were made by @Xinyi in summer 2020. In November, we had completed V1 design, which was made accessible via our public beta. In that process, we collected extensive feedback through user testing sessions and our public Bug / Feature Request page. Iterating on that feedback, we completed design of V2 in January 2021, and will be officially launching in the App Store sometime in February 2021.

Captivate with Onboarding

User testing has repeatedly shown us that many of our users ignore the most important feature of Lockcard: being able to interact with notifications. As a result, those users perceived Lockcard as a simple dictionary, as opposed to a workspace and memorization system that adapts based on your answers to notifications. 

For Lockcard V2, we decided to solve for that problem through better education by redesigning the onboarding experience. 

Lockcard App Onboarding
V2 - Onboarding

Enable to Search Fast

The foundation to a good user experience is delivering what the user needs as quickly as possible. That's why we focus on eliminating any friction when we search for words, and to embed Lockcard into the everyday life of our users. By reducing the amount of effort needed to learn, we hope to increase daily app sessions per user.

explainer popup on homepage - Lockcard

Reduce Cognitive Load

Traditional dictionaries tend to push a lot of visual cues and text all at once, and offer a lot of what our users label as 'unnecessary' information; types of English, origins of a word, related phrases...

We know that searching is very contextual, and that users need to approximate a word as quickly as possible to respond to a situation they are in. Searching a word always precedes a more urgent and important task. That's why Lockcard is designed to display only the information that matters the most. Any secondary information can be accessed via an additional press.

Dictionary Lockcard vs Merriam Webster
Lockcard vs Merriam Webster

Train to Recall (with Interactive Notifications)

We use notifications to flash searched words back to users’ phones, following a Spaced Repetition Schedule based on ACT-R, a modelling system introduced by Anderson & Lebiere that maximizes learning and retention. The notifications are uniquely interactive; your answers (Easy, Medium, Difficult) make the algorithm smarter which in turn optimizes your notification schedule.

Interactive Notifications - Lockcard
Lockcard's unique notification system

Organize and Measure Progress

When we encounter new words, we lack a centralized method for organizing them (printed notes vs note taking app vs dictionary). We end up losing track of what we have and have not reviewed, creating only more frustration in the learning journey. Following the CRUD principle, our Progress page serves as a workspace for tracking, measuring and updating the words we are currently learning.

'In Progress' page

Increase Trust in our Method

User testing has shown us that some of our users question the efficacy of Lockcard's teaching method. Some of the questions that pointed to that idea were: 'How do you know if I actually mastered a word?', 'Why do I need to use notifications?', 'Is it proven I’ll learn with the repetition?'... 

Lockcard's Teaching Method
Our Teaching Method

App Development

September 2020 - November 2020

The development of the IOS app was outsourced to a team in Vietnam, which I managed closely on a daily basis. The beta version of Lockcard was ready after 11 rounds of Q&A with developers. We're also partly to blame for it since we changed the scope of some features halfway through. If you're interested in the Figma hand-off file, please reach out!

Beta Launch 🎉

November 10, 2020

We decided to open our Beta to the public on November 10th 2020, and got noticed on AppAirport pretty quickly. The feedback has been overwhelmingly positive with the founder of @AppAirport Jordan Singer commenting over Twitter:

Now, Marketing

November 2020 - May 2021

In addition to our blog, which we've been keeping at pretty consistently, Lockcard uses its main Twitter account (@LockcardApp) to connect with more tech-minded people who see value in design and non-intrusive applications. 

Our sister Twitter account (@UnlockCard) mimics the use of our app using tweets to teach English words everyday, and encouraging people to comment several times to train their brain into 'active recall'. This is a temporary solution for our Android users out there who have to wait a little longer until we build it.

Cheeky brand pattern for @LockcardApp
Complimentary pattern for @UnlockCard

The plan is to continue to raise awareness over Twitter and our blog to get more users on the beta, collect feedback, iterate and finally launch public V1 (with paid features maybe), and see how it goes from there!

If you're interested in this space or working with us, DM us over Twitter. We're always looking for passionate people to help.

Preparing to go public

More beta testing

For the last month, we've been testing out app in the hands of as many people as possible. 10% friends and 90% strangers found over AppAirport and Twitter. We've gotten a lot of good feedback through user interviews, and made some of it public via Kampsite. We are now iterating designs and planning to release the public version to the official App Store in February 2021!

A really early award

We won an award for best app design since @AppAiport's inception half a year ago. We were selected among a cohort of 626 other beta apps, which totalled 1.1 million app views. What makes me so happy is that Jordan, the founder of @AppAiport himself, uses Lockcard and commented very nice things of it. 

Public Launch

March 16th 2021

We collaborated with Rowland for my very first motion design project, celebrating our official launch in the AppStore. The artistic direction experience from castlelab came in handy here to finish this project in very little time, and to produce an ad of respectable quality. 

Early Growth

I don't remember exactly how we got so many downloads on the first week, it was all very exciting. We had accumulated over a few dozens signs-up from our website, but most of the users were on the TestFlight version of our app, and so naturally they switched to the official public version... I think.

Week 1 stats

Paying the bills

The app didn't blow up like you see happening everyday on Twitter; after a few days, the number of downloads dropped to 5-10 per day. And since Lockcard wasn't monetized, Xinyi and I both started separate full-time design contracts around March 2021, and Lockcard became less of a priority for a bit. We spent on avg 10hr a week, and we found a junior iOS developer, Kouzi (senior at uni, new job at WeChat), help us maintain the app and naturally replace the costly team of contractors we had originally hired to build out Lockcard V1.

Fresh talents

May 2021

Lockcard users were starting to request new features and complain about the way certain existing ones were designed. Lost in all this scattered information, undecided on how to prioritize user feedback, and upset about our personal work dynamic, Xinyi and I decided to bring on fresh faces to the team:

  • Suchen - user researcher
  • Yuxuan - graphic designer
  • Zoey - UX & strategy
  • Jewel - blog writer
  • Davis - iOS dev (replaced Kouzi)
  • Falah - marketing
Team (missing Zoey) - June 2021

Hiring Fails

'Marketer' left the room

The last hire in May was Xinyi's roommate, and our very good friend, to lead Lockcard's marketing. He doesn't have a lot of professional experience there but two things that made us want to hire him: he was very excited about working for a friend's startup, and he's a very networking-minded person. For example, he stalks people very well and isn't afraid to flaunt on LinkedIn when something good happens, whereas I feel like that is arrogant. On top of that, we were seeing each other every single day since he lived with Xinyi, so I thought I could use those moments to do stand-up.

But very quickly, red signs started to pile up. He worked very little in comparison to other members, came up with abstract goals (whereas marketing should be metric driven), and mostly importantly, he never took responsibility or initiative, thus relying on founders to follow-up with him. 

Most will think that since he's not paid, it doesn't really matter. However, I have high standards for people, especially on a project I believe in so much as Lockcard. It took me too long to realized that I was spending more time on training him than the value we were getting in return. So, on the 3rd team meeting he skipped, we let him go. PS: He's still a very close friend today.

'3 Developers' left the room

Now this part is more interesting because both parties are to blame. Over the course of 3-4 months, we saw three developers join and then leave us. The first one was Kouzi who knew his starting state to his new full-time job at WeChat, and so naturally he couldn't stay longer. Then, we had Davis who runs his own app-development company, and was helping us irregularly throughout the week; he was quite knowledgable about widgets, but ultimately had to leave us to focus on his business. Last was Davis, who worked full-time for another company and so would spend his entire weekends on Lockcard. He was more skilled, more accountable and spoke better English, but falling sick one weekend meant that we lost 2 weeks of development time; he left us in August, afraid that he couldn't handle all the work on time. 

It is worth mentioning that all three developers spoke limited English, and that I spoke even worse Mandarin, and so most of our conversation had to be written so that I could translate them in my own time. You can read more about what's it's like working with a multi-lingual and multi-cultural team right here; it has its pros but it certainly did not help with speed and sometimes led to cultural misunderstandings.

my solution: WeChat translate each message 1 by 1

Developers are a weird specimen. I've worked with many over the course of my ventures, and I'm just learning that development differs from other professions in that it's very hard to do part-time and well. There's constantly a backlog of items, and being able to only tackle a few at a time, or even worse half an item per sprint (e.g. Mickey), creates a lot of frustration for a startup. For blog writing, it's okay to take more time because it's an independent activity that doesn't affect other work much. 

And so, during development sprints I put on my manager hat and focused on execution, which caused our developers to feel rushed and worry about writing low-quality code. And to some extend I agree with that criticism; the Upwork freelancers wrote code in a language that wasn't scalable, and Mickey left us with some major bugs. But I sincerely believe that growing a startup requires experimentation, and thus rapid execution without necessarily consensus from the entire team or bug-free features at every release.

Lesson learnt: we hired our first full-time paid iOS developer in September 2021, and it's much faster than before. [I'll go into more details about the new developer a bit later].

Learning about users

May - June 2021

Medical Students - a potential market?

In our quest to better understand and identify the features that make Lockcard so special, we decided to run a short experiment marketing the app to a new audience. 

Before that, we had mostly advertised the app to non-native English speakers, and to some people who were reading a lot and needed a dictionary handy. One day, we talk to the Fluent team and they tell us that our app could be extremely helpful to students in medicine since they've got to retain lots of new vocabulary every term. And so, Suchen starts to interview a few to confirm that theory.

Medical students - User Research by Suchen

The conclusion is that medical students have a lot in common with the user persona we had originally designed Lockcard for. For example, they are vocabulary-heavy learners who rely on repetition, they often memorize definitions by heart, they are learning with the intention of retaining information in the long-term (e.g. on the job, 5 years later) and happen to be very familiar with flashcard tools, like Anki or Quizlet.[All of our research around medical students and their learning habits can be found openly on our Notion]

As a result, we thought wise to build a quick TestFlight version of Lockcard offering a medical vocabulary list, along with lists specialized for IELTS and TOEFL. It was meant to be a prototype since each list contained only 10 words, and we were hoping that users would then reach back out asking for more words. However, we were unable to gather testers for this version, and it was thus forgotten. Rip.

[September 2021 update: vocabulary lists are now the next major feature on the agenda, especially since we have large TikTokers ready to partner]

Current Users - who are they?

With time, we've been able to learn more about our current users and how they use Lockcard. Back in June, Suchen led a diary study everyday for 2 weeks with 5 of our users. 

Meeting Primus - a power user

Among them is Primus, a software engineer in Shenzhen (China) who plans on passing his TOELF test later this year. He uses Lockcard every single day, and “feels excited every time he receives a notification”. He has 80 words ‘In Progress’ and 30 words ‘Completed’ on Lockcard, which is currently a record across our users that we know of.

Refining our value proposition

Since LockCard didn't have any sign-up process, or ways to identify unique user profiles, we were left a bit clueless about how to incorporate the feedback we were receiving across our different communication channels. Our Marketing was also all over the place; as a result of not knowing enough about our power users, we were using sharing different taglines, ads and messages when promoting the app. For example, to my Canadian friends I described LockCard as a better dictionary, but to my friends aborad still learning English as their second language, I'd describe LockCard as a tool for retaining new vocabulary.

Therefore, our design strategist Zoey led an internal design workshop meant to ideate and agree on a single 'How Might We' statement that would define our target audience, plan the short-term roadmap and make our online brand more cohesive.

the product team beginning Zoey's workshop

Building new features

June 2021 - Present

Unlimited notifications

One of the more requested features among our Chinese power users is the ability to receive more than 3 notifications a day. We were currently stuck on 3 because of 1) we didn't think users would want more in the current state of the app, 2) our code was written such that it was difficult to allow for more than 3. However, unlimited notifications was aligned with our refined value proposition, and so we tasked Davis with that; I believe he had to rewrite lots of foundational code to allow it. 


We were finally set to put a paywall on LockCard - for unlimited notifications - and measure real user interest. Our advisors at Velocity had indeed suggested all startups to monetize early to assess if one's product could in fact ever generate investor-level revenue in the future. For example, a $1 paywall (assuming embedded smartly) shouldn't deter all your users from using your app. In our case, putting a one-time fee to the unlimited notifications feature made sense. 

However, at the last minute, the product team decided to release the feature for free on the basis that our app was not attractive or complete enough to justify a price. 

I understand that we are curious about the willingness of our users to pay and are interested in getting some benefits. But in my own point of view, both of them are limited and we can speculate the result from our current data traction actually. Monetization is closely related to growing: Monetization is not the only way to check the value of Lockcard for our users, we can also get some from the number of download users and active users. Also, if the money that we can earn from this itself is still limited, why not be free for a while to give more chances to the potential users to take a look at Lockcard and have a try? I also think we need to prepare well before monetization, making sure that we offer a product with complete functions and a smooth experience. And if we growing fast after fixing bugs and launching new features, that will be a perfect time to monetize!

Apple E-Camp 2021

A dream come true, we were one of the ten apps worldwide to join the Apple Entrepreneurship Camp, which accepts teams led by a woman, and coaches them over the course of 2 weeks on several topics: UX design, new iOS features, PR, commercialization...

doesn't this look slick

Almost all of the content from this program remains confidential, but here are some of the biggest learnings for me:

  • Telling a story effectively requires enormous practice. The folks at Apple train their presenters weeks ahead to deliver a 5 minute long keynote. Every word is thoughtfully picked
  • ...


  • Our first call with an angel
  • Tool for Thoughts, and the fascinating community around memory and retrieval effect
  • Andy Matuschak
  • new team members
  • lockcard as a business?

The current team

The two most recent additions to the team are Pranav, a senior at high school who released his own education app, and Keval a Mechanical Engineer by education who's recently found a passion for iOS development.


Updated September 22, 2021 (but you can follow our daily adventures via Twitter)