Lingvist Course Wizard

Personalised langauge learning

👨‍🎨 Lead Product Designer + PM

⏳ January 2019 - June 2019

Course Wizard Product Hunt

Overview

More, more, more… our users wanted more content. They’ve already learned ~5,000 words in our general course and now they want to progress even further. This was a huge challenge for us, as creating courses is a manual process.

"If the course would expand or I am starting new course, I would probably buy new subscription as I was satisfied with Lingvist."

This wasn’t only a frequent request from users, we knew it was hurting our revenue, too. Our yearly subscription rate was no way near forecast and one of the top choices on an automated cancellation survey I created was “I’ve finished the course already”.

Course Wizard could help us solve this by allowing users to quickly create and customise their own courses. We also believe learning courses relevant to you will increase your learning speed which is our main company goal.

Process

We needed to validate a big assumption that our users would want to create courses themselves. To quickly do this we looked at creating an MVP on Android to launch, test and measure results. We could quickly iterate once we knew we had something that users would value. We would leverage the backend built for the Educators version on our Web app -- see Classrooms case study.

User flow

User flow

It’s always tricky deciding what’s enough for an MVP, especially since we had already released the Course Wizard for Educators which had lots of features. I created a User flow using Whimsical looking at the big picture and deciding on what exactly we needed to build to release as early as possible - think big and start small approach. With the user flow we could agree what the essentials were and what we would release later.

My main aim was to make the tool easy to use for all users but comprehensive for power users, who really want to dig into the finer details for added course personalisation.

We agreed on the initial flow which allowed users to input text by typing or pasting, see results - 50 words, view the sentences they would learn and then save it. Very quickly, in future sprints, we designed, built and released - deleting words, changing sentences, improved the publishing audio speed, error messaging, updated copy, naming pre-fill, voice input, onboarding and a potential growth loop.

Wireframes

While on a flight to Tallinn for work, I created wireframes to discuss with the team when I arrived in the office. I already could tell that we could use a lot of existing elements in our design system to build this flow quickly.

Wireframes

UI Design

Once we discussed the wireframes, I quickly refined the screens using our design components in Sketch and prepared the handoff for our Android developer. These existing elements would ensure consistency within our product, users would recognise the patterns and reduce development time. This would help us answer our assumption without wasting any time.

UI flow
Initial flow we released to small cohort of users.

Qualitative research

Throughout the whole process, I was guerrilla testing with colleagues in the office and anyone else who would spare a couple of minutes. I wanted to try it out on some existing users too so we invited them into our London office. As I’ve found, recruiting participants, especially for in-person sessions, is the most time consuming part.

We knew from looking at our analytics that there were certain places in the flow that users dropped off - at the start and right at the end before saving. We knew where users were dropping off but we didn’t know why. During the usability testing sessions, I asked the participants what “Publish” (which we used due to the audio files publishing in the background) meant to them, as this was the title of our last screen. 3 out of 3 participants all had concerns about what might be published - would it be shared on their social media or other places? This clearly had other connotations for users so we reviewed the copy and re-localised everything before the full release.

I also created a survey using Typeform to be sent to users via email once they’d created a course. We wanted to gather feedback on first impressions and relevance of content. We also asked what could we do to improve the feature, how would they feel if Course Wizard were discontinued and an NPS score.

The results were mainly positive and gave us constant guidance on what feature we should be working on next.

Some takeaways:

🚀

End of funnel improved

Increased from 14 to 64%

🏎

Publishing improved

Average publishing speed ~3 seconds

🌡

NPS = 19

Our base line to improve

Growth Loops

We assume the 1% rule will apply to Course Wizard with the 1% creating a lot of courses and the majority of users learning from these courses. This will hopefully create a lot of organic acquisition for us. I also spotted an opportunity for a potential growth loop within the product. For every course created by Course Wizard, we would allow any users share it when the course was completed. This would mean any users could join Lingvist and learn someone else’s course, then share it if they liked, and so on.

I created another Flow chart with my trusty friend, Whimsical. At the onset, it looked like a simple task to allow users to share via their phones native sharing functionality but when I started to look at the other side, when users would be accepting, things became quite tricky.

User flow

Monetisation

There has been a lot of discussion around where Course Wizard should sit in our product, once we were satisfied there was product market fit and when we remove the Beta label.

I created a chart with 5 arguments in preparation of a stakeholder meeting. This helped move the meeting along with digestible points for each argument and removed any gut feelings. We could just focus on what was the right decision for our users and to help achieve our main business goal - LTV/CAC > 2.

Paywall chart

Course Wizard demo

Unedited video of Course Wizard live on iPhone.

Steps taken to create a course in ~1 minute:

  1. Create - scan text I would like to understand
  2. Adapt - delete words and changing course size
  3. Customise - change a sentence
  4. Save - edit name and allow others to learn
  5. Learn - start learning

Conclusion

This experience has been, by far, the most challenging and illuminating I’ve had in my career. I was trusted with the huge task of delivering this brand new feature, as Lead Designer and PM. I had to work with our CTO, Head of Learning Experience and Computational Linguist on a daily basis, while liaising with our Growth, Support and Localisation teams. It was tough going at times but I would do it all over again if I had to.

"I couldn't believe that a program could accurately create relevant 'content' so rapidly, but it did!"

I gained so much experience that will be invaluable to me for the future and we delivered on an initiative which launched on Product Hunt in June.

We learned a lot and moved quickly working in a small cross-functional team. The manner in which it was delivered and the basis of a game changing feature could be pivotal for Lingvist.

Find out where Course Wizard all began