Free your data – data democratization at Telefónica.

Laura Velikonja, a data scientist at Telefónica Deutschland GmbH & Co. OHG, gave a presentation at Data Festival 2018 on the data democratization strategy being employed at her company.

“All employees should be able to access the pool of data collected by the company”

This was the prospect CEO Markus Haas opened up for his company about a year ago. However, this comes with very far-reaching effects and can trigger different thoughts and reactions, ranging from enthusiasm to skepticism. For example, this proposition raises the question: “Should all employees now learn to perform their own analyses?”. The issue of data protection also seems to be quite relevant in this context.

This article and the accompanying video explain how such a demand can be implemented in practice (i.e. cultural change) – but first it is important to give a brief insight into the data situation at Telefónica.

Data at Telefónica.

Close to 50 million mobile connections were made at Telefónica in 2017, producing a considerable amount of data.

There is a variety of data categories at Telefónica – ranging from internal data, such as data relating to sales or customer satisfaction, to external data generated by social media, for example.

Particularly noteworthy is the amount of network events generated by the mobile company. Network events are data records that are created from calls, WhatsApp messages, or SMS, for example. These network events add up to a daily workload of 50 billion datasets.

It is well known that data is compared to oil, with the particular common denominator of these two resources being that they only gain in value through refining or processing.

From traditional analyses to advanced analytics.

Similar to many other companies, Telefónica initially conducted predominantly descriptive analyses of small amounts of structured data – typical of an enterprise with a classic business intelligence background.

Due to the ever-increasing amount and complexity of data, these analyses evolved towards advanced analytics, that is, predictive and proactive analyses of large volumes of data, sometimes including unstructured data.

The next logical step in these developments was to make these insights generated by advanced analytics accessible to the business departments as well, thus putting the impact of data insights at the heart of the company.

Implementing the data democratization project.

In order to actually make the CEO’s goal a reality, a platform was implemented – the so-called Analytical Insights Center (AIC). This platform gives employees access to all analyses and dashboards of the Business Analytics & AI team at Telefónica. This means that everyone has access to properly prepared data and that more data-based, more efficient decisions can be made.

The question now is: how do you set the right incentives for employees so that they actually use the AIC in their daily work? Laura Velikonja presented four key factors:

The data quality in the AIC must be guaranteed at all times so as not to lose the trust of the business departments that use this data.

  • Marketing

In order to sell the AIC internally and thus increase use of the platform, various measures were taken at the company. For example, under the slogan “Telefónica knows”, articles about the AIC were published on the Intranet, a monthly newsletter presented new dashboards and analyses, and a screen in the foyer as well as a corporate video of the Analytics team encouraged employees to give their due attention.

  • Community and training

Of course, employees must also receive appropriate training in order to use the AIC in a meaningful manner. There are three different categories of employer training that require an increasing amount of data expertise:

  • Data worker: these employees are trained to find information in the AIC and interpret it correctly. They use an (internal) search engine for this.
  • Citizen data scientist: these employees from the business departments are data-oriented and have volunteered to be more involved in using the AIC.  To do this, they are trained in and use Tableau, for example.
  • Data Scientist: the data scientists themselves have direct access to the company’s data lake.
  • Access

The last important step in this data democratization strategy was to overcome any technical hurdles that could affect access to the AIC.

Development of the AIC and resulting implications.

In the course of its use, the AIC grew extremely quickly and became very complex. A number of measures have been taken to counteract this.

The first measure was vertical integration. Since it was sometimes difficult to find relevant information, the platform needed to be simplified. Three aspects were taken into consideration for this simplification:

  • Enablement of searches by subject area
  • Establishment of an exchange forum
  • Internal use case: implementation of a chatbot to provide support

However, these new measures still did not always help to optimally satisfy all customer needs.

In order to clear these problems out of the way as well, a horizontal expansion was sought. A new platform was created for this purpose: the Digital Data Analytics Platform (DDAP). Ultimately, this is nothing more than a Tableau server that offers good customization options.

The DDAP comprises three elements:

  • AIC Open Data à “Democratize knowledge”

Here, the raw data is available. The access to this data is controlled according to the data protection policy . Citizen data scientists can carry out their own analyses in this area.

In this area, the Analytics team provides the business departments with customized analyses. These are also access-controlled and can be published.

  • Insights of the business units à “Democratize data science”

Individual business departments are provided with access to the Tableau server – here they can use their own data or create visualizations independently.

Thus, according to Laura Velikonja, it can be summarized that internal data usage has evolved, in a figurative sense, from a “simple webshop” to a complex marketplace for analyses. This has caused not only the data, but also the analysis options, to be democratized. That is, the analysis options have been handed over to the business department and are therefore now at the heart of the company.

Conclusion – what have all these efforts achieved?

Reflecting on the developments and measures taken, the question of their effectivity arises. In her summary, Laura Velikonja differentiated between the AIC and the DDAP.

Results of data democratization through the AIC:

  • 3,500 AIC users are working with the information provided in the AIC.
  • 50 data workers have been trained.
  • More than 60 citizen data scientists have received training.

Results of data democratization through the DDAP:

  • 3,500 users also use the Analytics Suite.
  • 11 business insights with more than 800 users, from the controlling or B2B department, for example, are used.
    • Not all departments are connected here – but this is not the goal. Simply reaching a critical mass that is interested in analytics is enough.
    • “The gut feeling has been digitalized”. This means that formerly intuitive decisions can now be made data-based.

Overall, it can therefore be said that data democratization has made enormous progress at Telefónica.


The full-lentgh video of Laura Velikonja’s presentation is available here..