Last year I was part of a team — with Yemisi Akinbobola and  Ogechi Ekeanyawu — that won a CNN MultiChoice African Journalist of the Year award for an investigation into Nigerian football agents. The project, funded by Journalismfund.eu, and also available in an immersive longform version, combined data journalism and networked production with on-the-ground reporting. Here are some of the lessons we drew from the project…

The story you start with isn’t always the story you end up with

Often in journalism you set out to report one story and end up finding many others. You have to be determined enough to get the information you are after, but open minded enough to be prepared to adapt to changing circumstances – and meet your deadline.

That was the case with our investigation into the world of football agents and Nigerian footballers.

The initial idea was inspired by a discussion about the idea of wealth extraction and Africa: that those countries are not ‘poor’ but rather the wealth generated there is taken out of the country and held in assets overseas.

This story is often told around natural resources such as oil – but footballers, and their agents, could provide a different way of looking at the same issues.

In the end we stumbled across a much more striking story of exploitation, highlighting a need for football authorities and clubs to take more responsibility for regulating and undertaking due diligence on transfers involving African players. Along the way we had taken the first steps to untangling the wider story of the world of football agents as a whole.

Step 1: Map the territory

One of the great advantages of data journalism techniques is the fact that they allow you to get an overview of a large amount of information very quickly. While much is made of data analysis techniques for spotting stories, it is also an important way of cutting down the time required to map the field you are looking at, and identify the places where you might focus your efforts.

In this case we used scraping techniques – the process of getting a computer to extract information from many online sources – to draw an outline of football agents representing Nigerian footballers.

The site Transfermarkt listed players and their associated agents. It also listed further details about some agencies such as their location and key staff. These were spread across hundreds of pages, but an automated scraper was able to go to all of those pages and put the information into one document: a spreadsheet listing any player with Nigerian nationality, and who their agent was.

It was important to recognise the limitations of this approach: data may be out of date, incomplete or inaccurate. And so we would not be relying upon those facts but instead using the broader patterns to delve deeper and establish facts closer to their source.

What emerged was a picture of a fragmented market – at least as far as Nigerian players were concerned. Over 200 agents were linked to around 450 players. And for a further 50 players there was either no agent listed, or simply the description ‘family member’. But we could identify the agents who were most dominant, in relative terms, and where our work could start.

In addition to this dataset we collected other, complementary, data. When FIFA maintained a register of agents this included agents based in Nigeria and elsewhere. This list – although a few years out of date – was also used.

And the website Fieldoo also listed football agents who stated which countries they operated in: we collated a list of those who said they operated in Nigeria.

An ideal situation would have been to use FIFA’s own Transfer Matching System (TMS). Although this does not list the actual agent its ‘data dictionary’ (a list of the fields the database used) was important in giving an indication to information which is stored and elements which might be overlooked such as the solidarity contribution, training compensation, and jargon such as “Club intermediary commission” which would lead us to experts such as Daniel Geey.

Similarly we needed to understand the background to transfers. For example FIFA publishes its regulations on the status and transfer of players (PDF) and FA publishes its own regulations.

Alongside this ‘shortlist’ of agents we also had an insight into other features of this world: by scraping details on transfers both from Transfermarkt and Soccerway we could identify which countries players tended to move between, and which clubs featured most often. We could look at recorded transfer fees and how much money transfers involving Nigerian players broadly involved. We could look at their ages and how large a proportion achieved a measure of success in the top 5 leagues (England, Spain, Italy, Germany and France).

Step 2: Drill down, working collaboratively

Once we had identified the agents the next step was to try to dig further into their connections, companies, and ‘ultimate ownership’. There were a number of barriers to this: firstly, there were agents with no obvious company. Then there were agents whose companies were registered in countries with no public register. There were those whose companies were dissolved or dormant. And there were those with more than one company.

Data journalists Leila Haddou and David Blood were enlisted to help try to trace company details for the agents we had identified. This helped us identify any problems ahead of a bigger attempt at the same.

The first port of call for tracing companies was OpenCorporates, but we also drew on the Investigative Dashboard, Compass by Arachnys, Companies House and the various equivalents across Europe, Passport, Marketline Advantage, and for Nigeria in particular, we looked at the Corporate Affairs Commission.

Meanwhile, Oge Ekeanyanwu chased the human stories in Nigeria. While there were numerous stories of young players who had been duped by fake agents, convincing them to tell their story was a hard task.

The first source came through the help of a fellow journalist who knew a few players who had been victims of a fake agent. Still, it took a couple of weeks to earn their trust.

The biggest obstacle though was getting a statement from authorities (one lesson we learned was that authorities in Nigeria are not easily accessible). The spokesperson of the Economic And Financial Crimes Commission (EFCC), despite all efforts failed to offer anything beyond promises to find out what the commission knew about player trafficking.

This was the same story with the Nigerian Football Federation (NFF). It was only after numerous calls, and two cancelled interviews, that a telephone interview was finally held with the general secretary.

Step 3: Organise a hackday for a big push

For a final ‘push’ to dig details on the agents, we organised a hackday to bring together a number of parties. This was to be held at Birmingham City University with students from the MA in Online Journalism (since replaced by the MAs in Data Journalism/Multiplatform and Mobile Journalism) and the undergraduate journalism course; local open data advocates and members of the Birmingham chapter of Hacks/Hackers; and members of the BBC and its data unit based in Birmingham.

The open data project OpenCorporates was an obvious fit for the event: their #FlashHacks hackdays are focused on bringing volunteers together to map company connections. We contacted them with a view to making this hackday a #FlashHacks event with a focus on football agents.

The Tax Justice Network’s The Offshore Game project had also systematically catalogued ultimate company ownership in English football clubs: the man behind the project, George Turner, was invited to help.

It was important that the hackday was designed in a way that the widest possible range of people could participate: participants were presented with various options, from simple search and data entry to using company registers and creating network visualisations.

At the same time presentations provided extra support and guidance at different points: early on the participants were introduced to a shared Google spreadsheet. Later Hera Hussain explained the world of corporate registries and the differing levels of access.

Audio from Oge’s interview with a young footballer was played to help put all the work into a human context.

George Turner, OpenCorporates’ Hera Hussain, Leila Haddou and David Blood, along with Yemisi Akinbobola and I, provided a vital backbone of support in helping hackday participants access a range of sources to dig into company ownership and individuals.

Having speakers of a number of languages in the room helped: Russian, French, Romanian and Spanish were all needed to understand specific company registries and agents’ details on different pages, while in other cases Google Translate helped clarify details.

One web developer mapped the networks of players and clubs, while another identified the role of football academies in Nigeria.

By the end of the hackday we had traced company details for at least 40 agents and biographical details for dozens of others. But perhaps equally importantly, we had introduced journalists and journalism students to methods of investigation and a topic they would otherwise likely not be exposed to.

Step 4: Use visuals to narrow down the story

The biggest challenge following the hackday was managing the information generated. As is so often the case with data gathering and analysis, we had many leads – but needed to choose the story to focus our efforts on.

Oge’s work in Nigeria was now coming in, and it was clear that this human narrative would provide the spine of the report that would eventually emerge. To support this narrative the data would play two key roles: firstly the ‘bigger picture’ of the number of players involved, the countries they play in and the clubs who bear the responsibility for due diligence in transfers; and secondly, to illustrate the complex network of football agents.

David Blood, who had identified one of the most interesting agent’s networks, agreed to create a network diagram of that. Meanwhile Caroline Beavon was commissioned to work on the ‘big picture’ infographic and data visualisation.

It was during editorial discussions with Caroline to determine the visualisation that we arrived at a key insight: this story was no longer primarily about tax, but about agents and young players. The visuals would focus on the journeys of the players and the networks of the agents. Tax would play a bit part — for now.

But this story did open a door: through the experiences of young footballers in Nigeria we could shine a light on the world that agents operated in and the curious way that money moved within that.

It raised questions not only about the operations of agents but of those they work with, particularly football clubs. It established useful practice and areas for further investigation, and helped to establish a network of people who could help explore those leads, with two journalists expressing an interest in doing so, and the beginnings of discussions with news organisations in 4 other countries who wished to take up the story and follow it outside the UK.

A version of this first appeared on IQ4News in 2016.