From 1 to 100K users in 48 hours 🚀

Jean-Sébastien Wallez
4 min readJan 9, 2017


If you’re over 18 years old, you probably haven’t met Peter yet.

“Call me back when you have some traction”

When you launch a startup, the most common sentence you’ll hear from an investor is “Looks interesting — call me back when you have some traction.” I heard it all the time with my first startup, Noospher, and I always left the meeting wondering, “Well, how do I know when we have traction?”

Meet Peter 🙈

Peter’s a tutoring bot available through Facebook Messenger, always there and ready to help students with whatever questions they may have.

With Jérémy and Stephen, we launched Peter on November 4th 2016. Enchanté ! 😊

Peter’s powered by a unique combination of human and artificial intelligence.

Peter’s super easy to use: just send a question, wait a few seconds and you’ve got the help you need from a qualified tutor. It really works like magic! ✨

Our early users were pumped!

Marketplace equilibrium ⚖️

Everything was great but we needed to find our marketplace equilibrium.

We actually got an early peak of users about a month after launch. When we saw that, we opened up the service to a bunch more tutors — and that meant that we didn’t have time to check them all out, pedagogically speaking…and all of a sudden, our users were less pumped about the experience.

This was an URGENT issue 🚨 If students didn’t like the help they got with Peter, no more 5-star ratings, no more shares, no more talking about it with everyone they knew, and of course, no satisfaction for them or for us.

Gilles suggesting us to put a queue to take control back of our user growth and experience

The queue jumper hack 🚀

What is this? This is basically a waiting list that the user can jump by sharing Peter with others. Anyone can sign up for Peter, but they’re then put on a waiting list. The fastest way to jump up that list is by sharing Peter with your friends 🙊

We put this system in place over the winter holidays, when Peter’s activity was at its lowest. At that time we had about 800 students and 300 tutors.

On the first day that the Queue-Jumper was deployed, we got 16 new users. Not bad! On the second day, we got 64. And that’s when I started to think and called up Stephen and Jérémy, saying:

“Guys, if we can keep doubling every day, or even just every week, we’ll be up to 10,000 users really fast!”

But Peter didn’t double every day. On Day 3, he absolutely exploded 💥

Houston, we have a problem! 👨‍🚀

As you can see on the graph, we had 40,000 new students sign up IN. A. DAY.

We didn’t expect that — after all, we’d never seen that type of jump before.

An amazing team 💪

That’s when the questions about “traction” came together with something that Oussama told us one day about product-market fit:

“Success is obvious”

For us, success meant constant server crashes, upgrading capacities every 12 hours, and not sleeping at all over the holidays. But as a team, we came together, got the support we needed and overcame every issue that came up. And it just keeps going, and we’re grateful for it.

Peter is the greatest Christmas present we ever could have wished for. 2017 is turning into a very interesting year 🎉

A special thanks to Erika Batista, Fanny-Lou Benoit, Kyle Hall, Gilles Barbier, Serge Alleyne, Balthazar de Lavergne, Oussama Ammar, Willy Braun, Maxime Berthelot, Tom Benattar, Laurent Allard, the Amazon and Facebook teams for their amazing support, help and reactivity! ❤️

And if you’re interested in a rocketing AI ed-tech solution, whether as a developer or investor, give me a shout at 🚀