Things learned in 5 years of Journalism++

In early April 2017, together with Pierre Romera and Anne-Lise Bouyer, we decided to shut down Journalism++ Paris/Berlin, the datajournalism agency we created over five years earlier (Journalism++ Porto and Stockholm, our sister companies, are not affected by this change).

We created the company in 2011. It feels like another era. Wikileaks had just shown how large data dumps could change the techniques and processes of investigative journalism. Open data was blossoming everywhere. In politics, Sarkozy and Barroso were on their way out, the Poles had just given a new majority to Civic Platform and the only pressing question was whether the Euro would survive the winter. Good old days, in retrospect. In this environment, we set out to do three things: Build a European company, improve journalism (“Journalism++” literally means “journalism = journalism + 1” in most programming languages) and create a sustainable business. It was crazy ambitious and, back then, I didn’t think we would survive more than three years.

Build a European company

Our largest and least visible challenge was to build a European company. In 2011, I was living in Berlin and Pierre was living in London (he moved to Paris soon after). We decided to set up the company in Paris because it was easiest1 and thought that, since there was a single market throughout the European Union, managing a business based in another country wouldn’t be too hard. How. Wrong. We. Were.

We learned that the single market was for large corporations only. When we asked our French tax office how we could sort issues such as corporate tax and VAT, they told us that they had a special bureau for European companies, but they only received businesses that grossed over 300 millions a year. We were just 299,800,000€ short. We asked all possible “Do business in the EU” help desks we could find how we should organize. No one had a clue how we could run a micro-company across borders. “Just create a subsidiary”, some said. But, on top of doubling overhead costs such as accounting, it would have required the mother company and the subsidiary to invoice each other when working on joint projects. Since we were doing every project together, it would have created a horrible mess. Simply put, there is no legal solution to create a business unit across borders.2 Each euro of added value must be anchored and taxed in a specific country. In the end, the French-German Chamber of Commerce gave us the best advice: “Just give some money to each country’s tax authority”. Which we did. We created a fiscal vehicle in Germany to pay some taxes,3 and paid the rest in France. We still do not know if this was the good solution, since, when we asked the administration for an advanced tax ruling (a way to ask the tax authority for its opinion), they refused to comment.

In the end, being over two countries cost us a lot of money, took an extraordinary amount of time and added a lot of fiscal uncertainty. The extra costs incurred by lawyers and accountants was close to 2% of our gross income. The fact that we could not discount VAT from our purchases in Germany cost us another 1%. Over the lifetime of the company, that’s 30,000€, which amounts to the budget of one large investigation.

We wanted to be a real European company because we wanted to have real employees, not freelance contractors. Not only have employees more loyalty to the company than freelancers, it’s also much more comfortable, as an individual, when you don’t have to worry about your health insurance and payroll taxes. This worked, but the cost was high.

Finally, and perhaps most interesting if you’re thinking that what we faced were hurdles in the process of being removed by the European Commission, you should be aware that there is zero willingness to europeanize, anywhere. I already mentioned how clueless the help desks were. It goes further than that. The Commission, for instance, created the SEPA zone for payments, making - in theory - all payments within the eurozone the same, whether they are cross-border or not. None of our German counterparts (tax collectors and social security) accepted that we connect a French bank account to pay our dues. We had to manually send the money every month, which led to delays, which cost us around 500€ in total. Similarly, it’s impossible to pay for a business cell phone contract in Germany if the company is not registered there. We had to personally pay for our German SIM cards, another 500€ a year.4

To steal a metaphor from Martin Luther King, I feel like we were given a promissory note, a promise that the Commission and the Member States would help us build an ever closer union in Europe. But when we actually did, the check came back marked “insufficient funds” and we could only acknowledge the impossibility of living life as Europeans.5

Improve journalism

Before Journalism++, Pierre and I worked at OWNI, a French start-up for news that rose fast in 2010, winning an ONA Award for best non-English newsroom, and nosedived faster thereafter. When we left the wreck, we wanted to continue working together and applied at several newsrooms in Paris and London. At the time, there were no datajournalism units and no one agreed to hire us as a team. One would to take us, but I’d have had to work with the journalists and Pierre with the developers. No way, we said, and we created our own thing to do journalism with computers, regardless of job titles.

Today, a great many newsrooms in Europe have datajournalism teams, datajournalism is taught at dozens of universities and datajournalists win Pulitzer Prizes. Did we provoke this change? Certainly not. Did we contribute? Probably. We trained hundreds of professional journalists and students. We published the source code of hundreds of apps on Journalism++'s Github account. We spoke at dozens of conferences on every continent except Antartica.6 We worked with journalists and developers from all around Europe and took in a dozen employees and trainees. Some people told us that we inspired them and I’m very, very thankful for that.

But let’s not kid ourselves. When we started doing datajournalism ten years ago, we were considered young and innovative. In 2017, we’re much less young, but we’re still innovative. Something’s wrong with that. If a major change is to happen in an industry, it happens in less than a decade. Ten years is the time it took from cinema to go from the invention of the movie camera to the industrial complex we know today, for instance. Interestingly, media groups did radically transform other branches of their businesses, such as classifieds and printing – content production alone remained unaltered. This made me realize that most media groups did not care much about information in itself and were more interested in the social role they played. (I wrote about this at length in Celebrating Ten Years of Datajournalism.)

That datajournalism will remain a niche does not lower the quality of the amazing work done by datajournalists in Europe and beyond. But it was a sign for me that the time of pioneers had passed and that I needed to move on, even if our business model was solid – by industry standards.

Create a sustainable business

When we started Journalism++, we wanted to build paid apps based on data that newsrooms collected but were not exploiting to their full possibility. Such data included hospital and schools rankings, recipies, election results etc. We talked to half a dozen newsrooms in Paris, which either did not understand or declined because it was something they hadn’t done before. We carried out a test with one and, without pretending that our product was good, we realized that a newsroom would never be a good commercial partner. They had no idea how to build a conversion funnel (the process that starts with acquiring a prospective client and ends with her buying your product), going so far as advertising our product without posting a link to it.

From this first failure, we moved slowly towards becoming an agency for data-driven stories. By the standards of European journalism, we were very comfortable. Salaries were relatively high (for journalists, not computer scientists), were never paid late and we never cheated the tax authorities by hiring freelancers as full-time employees or having staff work on successive short-term contracts, a common practice. We built interactive visualizations for clients, mostly non-governmental organizations and think-tanks, we trained journalists and offered consulting missions and we applied for grants to work on purely journalistic projects.

Many grant-giving organizations we were in touch with7 pretended to be interested in helping us build useful tools. Instead, they wanted cool prototypes and did not care in the least if the product had any kind of usefulness or viability.8 This let us test crazy ideas, like Feowl, an SMS-monitoring solution for power cuts in Cameroon, or Rentswatch, a measure of rent prices harmonized across Europe in real time, but it never led to a significant breakthrough. There are two psychological approaches to seeing your projects rise and fall year after year. There’s self-delusion: You can pretend that it’s normal that all your projects fail and that, one day, you will have a really good idea and hit the jackpot. And there’s the cynical approach: You know that funders don’t care about what you do and you apply to grants with the same project over and over again, spending increasingly less time to produce your deliverables.

As I wrote above, we did not create Journalism++ for this routine. We wanted to create game-changing products and we failed, twice. The first time was in 2013. Like many others, we thought that network analysis was so cool that journalists needed tools to do it. We built, a network-analysis and data management platform, at the time when, Graph Commons, or Linkurious were starting. Building the tool before we did a proper market analysis was a bad idea. We folded it after less than 18 months of operations, losing about 80,000€ in the process, which forced us to fire four employees out of seven.

Our second attempt was jQuest, a platform for journalism schools to teach datajournalism and investigate collaboratively. This time, market research was good and the platform started well, with 30 schools in 15 countries and 500 students trying it. Despite the traction, we could not convince our funder to finance a second year of jQuest, which might have taken us to a point where we could have monetized it. That was in April 2017, at which point we decided to close the company.

I’m very proud of the five years I spent at Journalism++ and extremely thankful to Pierre and Anne-Lise, with whom work never felt like work. I’m also thankful to the hundreds of people who trusted us and helped us along the way. (I’m also thankful to the few who tried to break us: they made our resolve stronger.) But don’t believe that our experiment is proof that entrepreneurship in journalism is easy. Like the vast majority of founders, I’m white, male and wealthy.9 We didn’t start with a $100,000 gift from dad like Jeff Bezos,10 but having well-to-do families helped in the first years, when money was tight (Pierre lived in a flat owned by family and I once had to loan the company €10,000 to overcome a gap in cash flow).

Seeing close-up the contradictions of the European Union and of journalism let me think more clearly about a wide range of topics and for that, on top of the rest, was my time with Journalism++ worth every second. Now, after five years of not taking more than a few days off, I’ll spend some time reading, writing and meeting people – before starting a new project.


1. The French have a type of limited liability, stock company, the société par actions simplifiée, with no capital requirements.

2. Of course, you could argue that there is the European Limited Liability Company. Suffice to say that this specific legal form is itself different in each Member State to realize how useless it is for entrepreneurs.

3. A Selbstständige Zweigstelle, which gave us a fiscal number, with which we could hire under German law.

4. If anything, the roaming regulation made things worse for us. By letting providers add a fair use clause to their roaming provisions, the Commission destroyed the nascent market for real Europe-wide contracts. There were a few options before December 2016, there are none today, as all contracts are capped at 30 days of roaming per year.

5. Is the comparison with the what MLK described adequate? It’s certain that the situation of Europeans today cannot be compared with the lives of African Americans in the 1960’s. It’s also true that the lives of those who believed their rights as Europeans were foverer have been, for those who had their savings in Greek or Cypriot banks, or will be, for Brits living in Europe and European living in Britain, dramatically changed for the worse.

6. That’s taking the large definition of Oceania, though.

7. This does not apply to JournalismFund and JournalismGrants, which fund stories.

8. If they did, they would accept failures and research them, which they do not do. I wrote about this issue in Collaboration in journalism.

9. Read Entrepreneurs don’t have a special gene for risk—they come from families with money.

10. See this Quora answer.