Killing your babies: tales from the startup world

Erwan Lemonnier
4 min readNov 1, 2020

At the core, I am a software developer. Have been since the age of 12 when I got a pocket calculator with a 30 characters display, a few kilobytes of memory and a Basic implementation.

But while writing code has always been my comfort zone, I have long felt a craving for more, a longing to build a product, a team, a company culture. To become truly economically independent. So like many, when the time felt right, I took the jump and started my own company. And failed. And started again. And again.

This post is about the tough journey of trying to get a business idea off the ground. Then letting go off it.

Spotify adopted early on a generous option program. That’s all thanks to the entrepreneurial spirit of its founders, Daniel Ek and Martin Lorentzon: they knew that rewarding employees would ultimately create a favourable ecosystem for more startups to rise up.

In 2015, after 4 years at Spotify, my stock options started being worth enough that I felt ready to take a few years off and try my teeth at the startup journey. I wasn’t alone in making that jump: in fact, Stockholm is full of ex-unicorn employees turned entrepreneurs, largely thanks to option and incentive programs.

So I met

, ex-COO of Bukowskis, and we started a company together: Klue. A fully digital auction house, offering online valuation services, later pivoting to opening its own marketplace. Except that the traditional auction industry is a dying affair, and that bootstrapping a marketplace is immensely hard. So after 2 years of building various products and trying various approaches, we ended up mutating Klue into a Facebook community by merging with Lyxloppis, Sweden’s largest marketplace for second-hand luxury goods. In a way, it was a success: Lyxloppis is the biggest competitor to traditional auction houses in Sweden. But in many ways, it felt as a failure: all those products, discarded. And how do you monetize a Facebook community? Well, barely.

And here I was, a developer, with a bitbucket account full up with microservices and code for various MVPs, all obsolete because Facebook had become our technical platform.

It took me a while to let go.

It is a mental process to look at what you have done, at all the commitment you have invested in a company (and never before had I worked so hard for anyone!) and to step back and realise that there is no point slamming your head on that wall any longer. It takes time to realise the futility of going on. There is a lot of inertia involved.

It hurts to let go of your code, but that’s only part of it: a lot of the energy you invest in a startup is about telling yourself a story and making yourself believe in it. There is the story you tell investors, a story of a bright and rich future. There is the story you tell yourself, of who you are becoming (an entrepreneur!), of what you are doing (creating something unique!) and why you are doing it (the freedom! the empowerment!). Your startup venture becomes your identity. You start networking with new people: other entrepreneurs, investors. You acquire new skills at a pace like never before in your life: pitching, networking, accounting, pivoting, lean thinking, business analysis, marketing, growth hacking... Hell, you end up doing the work of 10 different persons in any normal-sized company.

You turn into wonder-woman. So of course, it is addictive.

But letting go is a necessity. It is the liberation you need in order to get the mental space to start anew.

Letting go of Klue wasn’t too hard for me. Probably because I didn’t truly believe in its core business idea.

Letting go of the code I had written was much harder. I had built this marketplace: (don’t bother googling it, it’s gone). I had a fleet of web crawlers sniffing marketplaces for luxury items sold by private persons anywhere in northern Europe. You wanted to buy a Louis Vuitton bag? If someone was selling one somewhere in Sweden, you would find it on Kluemarket. Not that I care for Louis Vuitton bags. I still keep suing up the old knapsack I had as a student, so I was definitely not my own customer target group. But the code was beautiful. A micro-service infrastructure, dockerised and deployed on Amazon Beanstalk, using AI to identify interesting items for sales, with high test coverage and a smooth deployment pipeline. It was beautiful. And the site had a lot of traffic. I felt proud.

I couldn’t let go of it.

So I bought back from Klue the rights to the codebase, rebranded kluemarket into bazardelux, refactored everything to make it even smoother and shinier, and kept it running for years. I had ambitions to grow its fleet of crawlers, support more and more sources, increase traffic and monetise it. But I had no time over to get it done.

Time passed, I started another company, got involved in other projects, and slowly but surely, what had been my most important occupation for a few years turned into a mild annoyance every month when I got the Amazon AWS bills.

Until one spring morning, years later, when I took a few minutes to delete bazardelux’s AWS account then walked out to enjoy the sun. In the end, it took only a few minutes to close it all down, but years to really let it go.

So was it a waste of time?


I learned a big lesson: how to kill my babies (that’s a figure of speech, duh.) And next time will be even easier. And equally liberating.

I also learned that in business, it does not matter how beautifully crafted your code, how smart your design and how smoothly running your infrastructure: if it does not serve a profitable business model, it is all wasted energy. Something that developers in larger companies are very seldom aware of.

And bazardelux? Well, I stopped paying for the servers, but I still had the code. So I published it on Github, here and here and here. And maybe it will help someone someday as an inspiration to solve some engineering problem.



Erwan Lemonnier

CTO. Fullstack developer turned entrepreneur. Ex-employee of Spotify and Trustly. Author of the PyMacaron microservice framework.