How we made the “Tuenti Mafia” article with Graphext
(Tuenti Mafia: así se gestó la mayor cantera de talento tecnológico español is an article published on Teknautas based on the data from more than 250 people + 30 interviews)
The Spanish version of this post, originally written by Analía Plaza
A Friday morning on late April, Victoriano and Miguel, from Graphext, met with Analía at La Bicicleta to have breakfast. Victoriano had contacted Analía some days before: Miguel and Victoriano were in Valencia, at Plug and Play, the startup accelerator, building Graphext. As we were in Madrid for a few days, we wanted to show Analía a demo. After ordering toasts and coffee, we powered our laptops and started digging into it.
Graphext is a tool to define and analyze contexts of information using public digital feeds. Observing people’s public profiles and how they interact with each other, Graphext gives insights about their relations and how they move information.
Looking at one of our contexts made up of Spanish tech entrepreneurs, investors and journalists, the second most shared Spanish media was El Confidencial. And the second most shared article of the year was one of Analía about Hawkers — How four young guys from Elche turn over more than 15 millions being hype experts.
Because of its innovation interest (El Confidencial Lab), Graphext is interested in El Confidencial, whose business model will be selling the tool to the media among others. And there was a story that could work pretty well: it implied studying the relations among many people — more than 253 employees from one company — from one sector on which El Confidencial is quite influential — business and internet.
The story was about the “Tuenti Mafia”, the Spanish homologous to the “Paypal Mafia”. Graphext proposed it days before to the El Confidencial’s director to publish it and to Analía to be the writer. Curiously, Analía had already thought about this story and started writing it a year before but stopped due to lack of time. She actually mailed Javier Escribano, the first person writing about the Tuenti Mafia, on May 2014 asking his permission to take the idea and make an article out of it (to which he replied back his approval).
Everything just fitted together. With Graphext they would gather and analyze data from the former and current Tuenti employees and by interviews with them the story would be strung together. We payed for the breakfast and went to work.
Gathering data
We had:
- A map with locations and names of ex-tuentis.
- Some lists of Twitter.
- Integrants of the Telegram group they share.
We had to create a context with all of them.
Contexts in Graphext are created syncing with twitter lists: each of us created our own and we began adding members. We went through all the names in the Telegram group. The twitter user name wasn’t always obvious: some search tricks were “name + tuenti + linkedin”, “name + github” (Github is a code repository where almost all developers have a profile whose user name is often the same at twitter) and even “name + dribble” (Dribble is a designers community).
Almost all of them had Linkedin but there were some without a Twitter account. We wrote down these profiles in a spreadsheet until we had a CSV importer ready.
The April 24th, when we started, Graphext only allowed adding the organizations (e.g. companies, universities…) someone was related to but not dates or roles. We improved Graphext to store this kind of information that was important for our article.
Selecting and arranging interviews
After several days of work on the data, some names emerged and we had to talk with them. We created another spreadsheet to keep track of every planned interview.
In brief:
- The first week Analía talked with founders and first employees.
- The second week, she talked with those who built startups afterwards (the original Tuenti Mafia).
- Since the beginning we knew that an important piece of the story were the ex-tuentis abroad. The third week was dedicated to them. The interviews mainly consisted on talking about their resumes. Every single one of them had been in interesting companies so knowing how they got to work there, how they moved to the next one and what they were up to in these companies (as well as commenting their time at Tuenti) was enough.
- The fourth week we tied loose ends and Analía spoke with some profiles that came up during other interviews or in the data. For example, it wasn’t until the end that it became apparent the importance of ex-tuentis in Spain. These stories needed some extra work.
There were 31 interviews including Adolfo, the waiter at the beginning of the article, without adding the conversations with people directly or indirectly related.
Many interviews were in person in Madrid — in offices or pubs — and some via Skype. Only two by means of cellphone calls and none on email. Some interview requests didn’t receive any answer.
We achieved a quite complete picture: people from management and customer care teams, designers and engineers; big startups founders and small startups founders; men, women, juniors, seniors, Spanish and foreigners.
Scheme and visualizations
Victoriano and Miguel considered several graphics (one, another) and finally opted for a Sankey diagram to visualize the flow of employees and their affiliation through organizations.
They created an algorithm to position the organizations in the diagram so that dates and number of employees were taken into account: the visualization represents automatically, and maybe more objectively, the time flow of employees through these companies. In order to include it in the article, they implemented the creation of embeddable widgets in Graphext; just by including an <iframe> in any webpage (as Youtube does with videos). This way the diagrams retain the interactivity outside Graphext (before it was only possible to export them as PNG or SVG) and metrics can be obtained.
At first, there was too much noise, but combining a few similar names (e.g. Tuenti / Tuenti Technologies), removing some (universities), and rising the minimum number of ex-tuentis on the organization (if there aren’t more than three, that organization is hidden), we got a nice diagram.
In the article we included a link to a more detailed version of the diagram.
So far, what caught our attention the most had been the flow of employees from big media companies to tech companies. The media crisis started in 2008 and Tuenti grew up in 2009.
Geographical information was also quite interesting. We exported the employees’ current location to CartoDB using the Graphext one click integration and customized the map there.
They are the experts on maps so they helped us. We can appreciate the employees flow — from Madrid to the rest of the world — and intensity — more opacity, more people — on the map we included in the article.
Adding and playing with more data we can obtain new visualizations and stories. How many female engineers has been in Tuenti? How many men worked in customer care? For how many of them was Tuenti their first job? How long did they last on each company? Did they stay for the same amount of time as in Tuenti (average time was of two years and a half to three years)?
It’s a matter of asking yourself questions. If you have any, about this or any other topic, suggest them.
Photos, text, layout
Photos: we knew since the very beginning we would be taking pictures, what we didn’t know was of whom. The photographer was Victoriano and he was in Valencia, so the photos had to be taken in two days, during his next visit to Madrid.
The data itself told us which teams to highlight, both in text and in pictures.
Text: with the scheme done and the interviews written down, it took Analía two days and a half to finish the article. Slowly but surely.
Layout: We had some nice references (one, another) but not everything was possible and we thought too much eye-candy might put the reader off.
Victoriano and Miguel in Valencia, Daniele at El Confidencial Lab, Sergio and Analía at Teknautas and a Slack channel to comment everything. We edited and did the layout with the usual template. We added an index to divide the sections and make it apparent it was a long story. We included a collage made of Instagram pictures of the employees to try to amuse the reader.
We didn’t achieve it with everyone, sure: it’s a long and niche topic. But the ‘feedback’ of those who read it through, close or not to the story, has been quite positive.
As for the context we were aiming to, it’s the third most shared story of the year so far, only surpassed by two news about acquisitions and investments, neither of them an article.
To sum up
On May 26th afternoon we finished and it got published the 27th at 5am: 4 weeks and a half of work and, on the edition, layout, visualizations and text, eight people had worked on it.
Overall it’s been a great experience. Analía usually complains about the lack of resources in digital media compared to printed press, where every article implies a team of photographer, infographist, editor and writer working together. Also about the technological limitations of some newsrooms and CMS. Working in tech startups, she’s seen quick developments and internal tools.
The article has been a perfect match between for both Analía, as a journalist, and Graphext.
- Graphext has implemented new features that will be proved useful again.
- Since its conception, it’s based on data. Sometimes we had the intuition and the data confirmed it (e.g. The Cocktail as a talent pool) and sometimes the other way around (e.g. ex-tuentis that stayed in Spain).
- Not a single email. Everything on Slack or Google Docs.
- It’s been cost-efficient, we didn’t halt our other lines of work. Victoriano and Miguel have followed their acceleration program and Analía did her duties with the other clients. The easy access to sources and the positive nature of the topic helped.
- Multidisciplinary team. Everyone does what he or she is best at and enjoys it much more.
From Tuenti Mafia have come agreements (we know that some potential clients have contacted some of these startups), an use case to explain Graphext to other potential clients and some ideas for new articles.
You can get in touch with us on Twitter (@graphext & @lalalalia) or by email (info@graphext.com & analia.plaza@gmail.com).
