The following is not a success story. Pretty atypical for an article topic but I thought it might be of some interest to show you the work of a failed startup. Before starting your own project it's always beneficial to get some feedback from others who tried and, not always, succeeded before you.

The NewCo project

Back in 2012-2013, four people, including myself, launched a startup project called NewCo, as in the New Company. The idea was to offer a place where consumers would exchange information about the products they use and love thus empowering you to make better and faster buying decisions.

NewCo empowers you to buy better

No prior experience in the web: four engineers from the same school, a very good one though, but with very little experience in what building a web startup would mean.

No traction: while I'm sure this space is still promising for new startups, in such a B2C configuration you would need a lot of traction to get things moving. We failed miserably at that, the product wasn't compelling enough to attract readers and contributors.

No money: we had the idea that once a user had the information he wanted before buying a product, we would just have to show him available resellers in order to get money through affiliate marketing. Well, this way, we made 30€ in a year.

Co-founders disagreements: trust me, don't wait one year to talk about who owns how much of the company. Maybe you don't have to incorporate right away but have at least a written agreement of some sort.

A working website: at least being engineers was useful for this.

The development was mostly done by AGASS007 and me (blaze33). The stack is old, mainly built around Django 1.4: Python 2.7, Heroku cedar-14 (upgraded from cedar-10), redis, postgresql, celery.

Advice to future entrepreneurs

  1. Technology matters as long as it doesn't get in the way of all the rest.
  2. Design, UX, marketing, communication matters a lot ! Don't call yourself NewCo, as in New Company, really ;)
  3. People don't read, much less contribute on the Internet. You can't build your success on asking them to do stuff for you. Build it around what they would naturally do with the right tool (and build that one).
  4. Choose your co-founders wisely. If your co-founder show up one day with another associate without having consulted you first, run ! Trust is everything.
  5. Making money is a necessary evil, I've got a rent to pay, food to buy. The sooner you can break even, the better.

Resuscitating the app

Long story short, the original website is still running and I wanted to replace a feedback form that no longer worked by some comment about the startup story. Problem: I could no longer run the website locally and my code push were rejected by heroku. Two days of rabbitholing later I managed to rebuild and deploy the app.

Some technical details: You know what happens when you don't touch a project for 4 years: it no longer runs!

All in all there was some work, mainly updating dependencies, without breaking the app, to have it run locally first.

Pushing to heroku no longer worked too as they updated their cedar-10 stack to cedar-14 (Ubuntu 10 vs Ubuntu 14 mainly).

How to upgrade from the Cedar-10 stack to the Cedar-14 stack.

Still pushing to heroku failed. Turns out I was using an old and no longer supported python version. Adding a runtime.txt with python-2.7.13 fixed it.

Another fun thing that can happen in 4 years is that some packages are no longer available on pypi, or not at the version you need. Chasing github repos and installing from them solved it.

Opensourcing the code

The only useful thing I could still do:


I worked one and a half year on this project when the other co-founders decided to go their own way and incorporated their own company without me but still reused all my code and simply told me to sue them if I wasn't happy.

Comes the lawyer who aptly counselled that it would cost more than I would ever get back considering they weren't making any money.

Still I'm grateful for this experience as it gave me a solid technical background that I was able to reuse by working for other more successful startups.

Thanks for reading !