Long live Peter ✌

It’s amazing what a team of passionate friends can achieve together when working 18/7 in a house (with a nice view to the sea) during 3 months.

In 90 days, we redesigned our product from scratch, talked to 10K+ users, developed a ‘cutting-edge’ software to map our code (that ended in the trash), tried to hack Instagram, changed our logo in 15 mins, worked with a dozen freelancers, tested a premium service for the U.S. market, closed 5869 support tickets, completely rethought our community building method, made our own Tinder-like app and swiped 30K+ profiles to ensure that we had a 100% students user base, wrote and scripted courses about Napoleon and other historic figures, learned to develop with Unity/ARKit and launched our first AR ed-app. By that time, Peter has received more than 1 million messages from 100K unique users but despite these nice numbers, it isn’t catching on.

Peter’s market place issue 🚩

The big discovery of Peter as a peer-to-peer learning market place has always been that there are students ready to help their school friends with their homework for free. Few people believed us when we demonstrated that.

But here is the problem with Peter’s model.

In blue, you can see the number of courses per day. In green the number of students that didn’t find a tutor. And in purple, the number of tutors that didn’t find a student to help.

This graph shows a classic week for Peter with the main user activity happening from Sunday evening to Wednesday. What you can see with the blue line is the stream of peer-to-peer courses, everything looks great here.

The green line shows the number of students that didn’t find a tutor that day. What’s really great is that the line stays super low meaning that almost 100% of our students can find a tutor to help them, even cooler Peter usually finds that student ready to help you whatever the subject in less than 1 minute! 👏

You guessed it. The problem comes with the purple line. That line shows for instance that on the 12th of December, 983 tutors didn’t find a student they could help for their homework… This market place disequilibrium is due to a lack of students asking for help to Peter and even when we drastically improved the quality of our tutors through selection, it’s still not taking off.

The three reasons why Peter isn’t working 🚀💥

1. Kids hate homework

In most countries, our children spend their childhood sitting on a chair in a closed and boring room being passively instructed and continuously tested.

Does that looks like an exciting way to start your life?

Five days a week, they are forced to stay up to 10 hours in a school building. Add on that transport time and what the education system elegantly called “homework”, which means: work to do at home… Instead of spending their small time left with their family, friends, exploring nature or playing games.

At Peter, we think that this education industrial dating system and its constraints has a huge impact on kids’ choices. Consequently, their curiosity and interests to learn new stuff outside of school is diminished.

School might be a old memory to you but try, just for a minute, to put yourself into our kids heads. If you would be forced to do something, even something that you truly love like eating your favorite meal, playing your favorite board game or even sleeping, for half of your day, what would you do of your time left when you’re finally free from that obligation? Probably something else.

Because the school system is forcing kids to “learn” 8 hours per day (and most of the time stuff that they don’t even care about)… When they are outside of school, kids naturally want to do something else than what the system told them is the right way to “learn”. Sounds logic right?

If we translate that principle as a market framework, anything that raises kids (5–17 yo) time of learning, they won’t buy (in case of a free product, buying means using). Anything that promises them to decrease their time of learning, they’ll buy. And anything that offers them more fun and entertainment, they will have way more chances to give it a try and if they like it, to buy it.

Their parents, usually work the opposite way, they are more incline to invest in products and services that increase their kids learning and working time.

Market framework to explain the paradox existing between kids and parents needs. This concern only primary, middle and high school students from western countries. Once in college, students become autonomous.

Of course there are some exceptions. By changing its vision of school time and homework, Finland now has one of the best education system in the world.

But excessive school hours and homework are still a reality in most western countries like France… And even if Peter is a fun product trying to help them enjoy a bit more that painful moment, we are not fulfilling an authentic need and others online solutions are doing far better than us to realize students urging desire to spend as little time as possible do their homework (c.f. 3).

2. Synchronous peer-learning isn’t 100% reliable (for kids)

This is something we worked a lot with Peter. How to make sure that when a user needs help on a subject, we connect him almost instantly with the best student available out there to help him debunk his problem?

The reality is that human interactions are complex and unpredictable, it’s simply impossible to identify all the factors occurring during a conversation. May be a new homework chat between two random users is gonna start by:

“Hey, are you Paul Martin? The one from my school?”

Because one of the users knows someone who’s name is “Paul”, he or she wonders if it is the same person. That simple exemple shows how a conversation that could have been focused on Maths, becomes off topic.
Even by using behavior’s driving techniques and involving Peter in the conversation, we can’t avoid such scenario because this is the natural way kids talk when they meet someone for the first time. The only “issue” for us is that the value proposition of our service isn’t delivered correctly there.

What’s crazy is that we actually did quiet well at orienting the subject of users’ chats through conversational design, we now feel like experts in that field.

We used words and concept detection to analyse and program when Peter must intervene in a conversation and what he has to say. Peter is now able to successfully handle subjects like dating, bullying or fun and act accordingly. This psychological tricks helped raise our user’s satisfaction rate to 63%. Considering our users’ age and the subject (homework), it’s not that bad!

But the only way to really solve that problem is to switch from a synchronous system like chat to an asynchronous one like a forum where you can have moderators that validate, correct and sometime enrich each answers.

3. Google is better to cheat on homework

During our journey and like explained above, we remarked that the demand was higher from students ready to help than from students asking for help, provoking a counter intuitive and unexpected market place disequilibrium.

This are the three main incentives for students to help others for free on Peter:
- Social gratification (homework are also a cool way to make new friends)
- Self-challenge (trying to solve problems without adult judgment is fun)
- Pass time (messaging is the number one activity for teens on smartphones)

On the other side, what middle and high school students want when looking for help for their homework is simple: they expect a fast and correct answer.

Students naturally tend to pick solutions that will help them decrease their “working time”. And to be honest, there is far better services than random students-to-students matching to do your homework in no time.

The number one is Google. You can actually find instantly the answer of more than 80% of the exercices we saw on Peter, just by typing the first sentence of your problem in your search bar and Google will redirect you to hundreds of homework forums. The most famous one is brainly.com

The advantage of homework’s solutions services like Brainly or Snapschool is that they are public forums. Once a student asked a question for an exercice and received an answer validated by moderators, the answer doesn’t vanish and becomes searchable by thousands of students. Other homework’s solutions like Socratic or Wolfram are even replacing human help by algorithms and give you step by step explanations. Pretty cool right?


As a team, working together for 3 months in the same house was one of the best decision we took. It could have take us a year or more to understand all that otherwise… We still believe that one day tutors powered by AI/AR will exist but we did’t find a path to get there. More technical teams like Socratic, Wolfram, Mindojo or Google are in a far better position to execute that vision.

Why aren’t we launching our own homework forum? With new technologies emerging like AI/AR/Blockchain, launching a forum in 2018 doesn’t excite us that much and we think that other teams are already doing a great job there.

What’s next for Peter? 👊

Peter’s story doesn’t end here! Our bank account is still full of cash and the team ambition didn’t change one iota. We are already working on pivot ideas.

  1. We’re looking for a conversational design solution to the “What should we talk about?” psychologic stress existing in friends’ conversations.
  2. A crypto-money for teens

Just for fun

Here’s my two cents about how to get rid of our obsolete industrial school system during the 21st century. One of the solutions could be to create an incentive learning system that would reward people for studying what they truly want. It could work through the creation of a student’s crypto money (giving people real purchasing power) that they would earn by reading or watching educational content online like Wikipedia’s articles or videos from the Khan Academy, and distribute tests and digital degrees that would certified students’ knowledges making them employable by the labor market.