How we crashed our EdTech startup

A mechanical monster in Navadra

The last 3 years have been both exciting and demanding.

Within this timeframe I converted a childhood dream into a startup called Navadra, found partners ready to take the ride, built together a team of 7 and eventually did not convert this into a viable company.

The last year was especially tough. We couldn’t afford to pay me a salary so I cumulated a full time job during the day and worked nights and weekends on my startup. I felt exhausted because of the few hours of sleep and frustrated by the little time I had available for my wife and friends. At the same time, I felt excited because we had one last chance to prove our business model and the whole team was fully dedicated to try this last shot… but it did not work either.

When starting this company, my objective was to have a positive impact on as many teenagers as possible. We failed since a lack of a sustainable business model prevented us from expanding Navadra’s impact.

When you dedicate so much time and energy into a project, failing is harsh.

Naturally you start wondering:

“Is there anything we could have done differently that would have changed the outcome?”

The point was not to live the rest of my life full of regrets but rather to learn as much as I could from this experience.

I had 2 objectives in mind when writing these articles:

1. Do a proper post-mortem on my experience to better learn from it.
2. Help future entrepreneurs avoid doing the same mistakes that I did.

Although the primary goal of these articles was not to tell the whole story of Navadra, I believe a bit of background information might be helpful to better understand the examples and stories I mention later. That’s why there are 2 articles:

1. The first article relates how we crashed Navadra
2. The second article lists 11 advice I would give to a future entrepreneur by reflecting on my own experience

If you prefer to skip the story of how we crashed Navadra, you can jump right in to the article listing the 11 advice.

Navadra used in classrooms

1. How to fail by trying to please 2 types of users with different expectations

Back when I was studying biology and engineering in college I loved giving math lessons. With one particular student, I had great troubles motivating her to learn and, one day, I decided to try a different approach: I developed a very basic game with VBA and Excel to learn multiplication tables. Within few weeks, she new all her tables and, more important, she started to regain her self confidence. That day, I realized that games could be an incredible leverage for learning. I also thought and still think that designing a game to help kids learn mathematics more effectively would be one of the most fulfilling job ever. This idea stuck somewhere in my head.

Six years later, in september 2015, I decided that it was time to leave my strategy consultant job and make this idea a reality.

I started the project alone and spent the first 5 months programming a MVP (minimum viable product) to start iterating with real users. In the meantime, I was looking for cofounders and met Jérémie (game designer and COO) and Julien (web programer and CTO) who ended up joining the adventure around the time the MVP was finished. Together we defined the mission of Navadra: “help middle school students regain confidence in their maths skills”.

During the following 6 months, we iterated first with middle schoolers then with math teachers. We managed to organize face to face demos in classrooms and the great success of these sessions pushed us to continue. We acquired our first 1,000 users with these sessions but we could not organize face to face demos in every classroom so we started thinking about how to scale this great acquisition lever. We ended up developing a specific teacher interface so that any math teacher could select exercises for their student and follow their progress.

During our iterations we were faced with the dilemma of either making the game as fun as possible with fast-paced mental calculation but little pedagogic value or choosing more scholar exercises which had better chances of having a positive impact on teenagers’ grades. We chose the more scholar exercises since we thought it was the only way to stay true to our mission.

After these iterations, our game looked like this:

You would start the game by creating and customizing your own character.

Then you would train to sharpen your mind and master the magic of Navadra with mathematics challenges. You would get to choose which school of magic (Fire, Water, Wind or Earth) you want to improve on, resulting in different math exercises (arithmetic, geometry, data representation, etc.).

You would then be able to learn new spells of various effects within that school of magic and upgrade them as you become more powerful.

Finally, you would be able reuse a mix of the notions you have learnt during fight against monsters in a time-constraint format.

You could even team up with your friends to fight epic monsters you could not defeat yourself.

Obviously there were a bunch of other features: global and grade-specific leaderboards, chat and messaging between players, achievements, honour titles, monthly seasons with prizes, progress reports, etc.

In short, Navadra was a social « Role-playing game » with a cooperative dimension where maths were at the core of the game.

Back then, we thought it would be possible to have a game that would please both:

• Teachers thanks to its high pedagogical value
• Students thanks to its highly engaging game mechanics

The problem was that these objectives were quite contradictory so you couldn’t really focus on one without negatively impacting the other.

That’s the first big mistake we made.

At this point we should have made a clear priority between these 2 objectives while being aware that:

• If we chose to focus on teachers’ objective, we would probably never find a business model that would allow us to scale our startup
• If we chose to focus on students’ objective, we would not really fulfill our mission which was to help them regain confidence in their maths skills

Retrospectively, it is still a tough choice to make. Back then we chose the worst option: not choosing explicitly between the 2.

2. How to fail at convincing investors

During the next 6 months, we went from 1,000 to 50,000 users without spending a single euro in marketing. Word of mouth between teachers was working.

So far our product was free of charge and our plan was to have parents (and not schools) pay for their children. By doing so, there was two big uncertainties:

1. Value proposition: would the product still get high engagement from teenagers if it was mainly used from home and not from school? Would their parents pay for it?
2. Engine of growth: would word of mouth between teenagers work as well as word of mouth between teachers ?

We tried answering the first question by testing a freemium business model designed to be used mainly at home. We had an average conversion rate from free to paying user but our user retention was not good enough when teachers were no more involved. We carried out face to face user interviews with middle-schoolers and understood that we needed 2 things if we wanted the product to be really used outside of classrooms:

1. We needed to transform our desktop app into a mobile app (for teenagers, accessing their phone was much easier than the family computer)
2. We needed to go much deeper into the “gaming” aspects to provide a really engaging experience (at home, teachers would be no more here to keep the teenagers engaged)

Back then, we were a bit naive to think that, since we had already a bit of traction, it was now time to raise funds to go faster and realize these 2 points. Turned out, French investors are a bit careful when dealing with Education startups, especially the ones targeting K-12 (kindergarten to 12th grade). We can’t really blame them, Education for K-12 is one of the hardest sector to innovate in France and there’s an entire graveyard full of well intended startups that tried. Anyway, we spent 4 months talking with various investors, had advanced discussions with some of them but, every time, it ended up failing at the last steps because we hadn’t proved our business model yet.

That was our second big mistake: trying to raise funds in a challenging K-12 sector without having proved our business model yet.

It was a great disappointment but we were determined to move forward anyway.

We organized a crowdfunding campaign in 3 weeks and successfully gathered 60k€. We used that money to hire talented people: 2 employees and 2 interns to help us realize our next version. We spent the next 6 months working on our mobile app and gaming experience with lots of user testing along the way.

3. How to fail by underestimating the impact of the context in which your product will be used

On January 2018, we released the first version of our new mobile game. We were pretty confident since all teenagers that tried our prototype were much more engaged than with the first version. A large majority asked how they could continue playing at home, others would stay an entire afternoon playing with a simple prototype.

When we released our new version, it looked like this:

You would still start by customizing your character, except there were much more options and characters were now 3D models.

As one of the most promising student of the Knowledge forces, you would be sent to explore Navadra and free it from the Ignorance forces.

You would meet non-playing characters that would give you important quests to achieve.

To fulfill these quests, you would have to fight the monsters sent by the Ignorance forces using your magic.

To be able to cast your spells, you would need to solve maths challenges. The difficulty of these challenges would be automatically adjusted depending on your previous answers (among other things).

If you had a sufficient number of challenges right, you would defeat the monsters and would be one step closer to freeing Navadra!

We really thought this new version would be a big hit since most teenager tests we made were a success… but it didn’t. We had much better retention and conversion to payment than before but our user base wouldn’t grow naturally like it did before.

We started speaking to teenagers to understand why. The issue was that word of mouth between teenagers was not working like we hoped for. Our users enjoyed very much the app and cleared the current content in no time so they naturally spoke about it to their friends… and that’s where the viral loop broke. Our users observed mostly the same reaction from their friends:

“Your math game looks fun but I’d rather play real games when I’m outside of school.”

Thus, their friend did not even try the app.

That was our third big mistake: we had underestimated how differently the app would be perceived depending on the teenager’s context. In classroom, playing Navadra was an incredible engaging alternative compared to traditional math lessons. At home, it simply couldn’t compete with other games the student had at their disposal.

This statement seems obvious but we thought that, at home, Navadra wouldn’t compete with other games but rather with doing traditional math exercises. Unfortunately, we understood that the main driver that pushed teenagers to do their math exercises at home was not to get better at math but simply to do what was requested of them by their teacher.

That’s also at this point that we understood that the value proposition of Navadra “get better at math while having fun” was too long term for teenagers. Most teenagers weren’t really interested in doing math training to get better on the long run but rather interested in having a better grade to their next test.

4. How to fail by underestimating the impact of cultural specificities

We were sorely disappointed but, once again, we decided that we shouldn’t stop there and try everything we could. We identified 2 alternatives at this stage:

• With the help of teachers, make Navadra compete against traditional math homework rather than with other fun games
• Make Navadra’s value proposition more short term by redesigning the game to be a “revision companion” focused on bringing better results to the next math test

Both alternatives would just be a way to acquire and gain teenagers’ loyalty. We were hoping that, once the teenagers had tried Navadra, they would keep using it for its fun aspects and eventually end up getting better at maths and improving their self confidence.

We decided to go for the first option which implied bringing teachers back in the loop. We chose this because we already had a 2,000 teachers community and also because, with my former private math teacher experience, I experimented myself the power of an engaging tool with the support and advice a teacher can give and I strongly believed Navadra needed teachers to be optimally used.

For the next two months, we developed what we did on our first web version: a dedicated teacher interface to allow them to create their classrooms in the game, select exercises for them and follow their progress. In parallel, we leveraged our most engaged teachers to help us create new exercises. With this “Teacher app”, instead of giving their students a traditional homework, teachers would configure a “mission” in a randomly generated dungeon that students would need to complete while doing math exercises. Clearly, it was a childhood dream becoming reality for the whole Navadra team!

We knew teachers were not going to pay for it so we made this teacher interface and all product usage in classroom entirely free. It’s only if teenagers wanted to continue playing at home that their parents would have to pay for it. This system could perfectly work in the Netherlands as Squla had demonstrated, why wouldn’t it work in France as well?

It did not work as we hoped for and for a simple reason: there was a fundamental culture difference between the Netherlands and France. In France, especially among teachers, there is this strong belief that Education should be 100% free. Therefore, even though the teachers could use Navadra for free in their classroom, the fact that students would have to pay if they wanted to continue playing at home was sufficient to prevent teachers from using our product all together.

That’s our fourth big mistake: we underestimated how radical French teachers would be about free Education compared to Dutch teachers.

We decided to try a last attempt in the quest to find our engine of growth. We tried to target teenagers’ parents with targeted press release, influencers and paid advertising.

We did all what we could with the little budget we had left but unfortunately, it wasn’t enough to sparkle the fire.

Understanding that we would quickly run out of cash, we even considered a buyout. We did meetings with major players in the French Education business. Unfortunately, these players were facing hard challenges in their historical industry and were trying to diversify themselves in other industries rather than investing even more in traditional Education.

Out of budget and energy, we finally decided to stop the adventure there.

Before closing the company, we uploaded the code base of the desktop version on Github with a tutorial to help every teacher setup Navadra for free in their school.

We are sincerely hoping that, even though we couldn’t transform this into a successful business, this free version of Navadra will continue to help teenagers build up their confidence in their math skills.

In this next article using the things I have learnt from this startup adventure, I summarize 11 advice I would give to future entrepreneurs.

What did you find the most interesting in this article? Is there anything you would have done differently?

If you liked it, please don’t hesitate to hit 👏 to help others discover it as well. Thanks!