Why do organizations want to become data-driven,and AI powered? by Richard Benjamins

By Guest Contributor Richard Benjamins

Author of A Data-Driven Company, Richard Benjamins, explains why organizations want to become data-driven and AI powered and why you should consider it too.

A decade ago, a McKinsey report – ‘Big data: The next frontier for innovation, competition, and productivity’ (Manyika et al., 2011) — brought big data to the attention of corporate boardrooms. Since then, many large enterprises have started their big data journey, as part of their broader digital transformation, to become a data-driven organization. The vision was that big data would help organizations optimize their core business, power new data-driven products and services, and even create entirely new businesses. Today, virtually every business process is run on an IT system, and therefore generates data. This data can be analysed and used for optimizing the process; it can be combined with other data sources to enable further optimization across the enterprise; or, it can be combined with external data sources to help generate even more new opportunities. More recently, the same happened with AI, and AI has overshadowed big data in terms of potential value and attention. Indeed, AI is considered one of today’s most promising technologies — it’s been compared to revolutionary game changers like electricity and the internet. In fact, many governments have enacted strategies to maximize its benefits, while ensuring that potentially negative side effects are understood and dealt with.

However, becoming data and AI-driven is not an easy task. It represents a long process that’s integral to the digital transformation all large organizations are going through. Figure 0.1 shows Telefónica’s data and AI journey, which has been long and complex, with many ups and downs. There are two main learnings from the undertaking:

The journey is a phased approach; it’s almost impossible to jump directly to the desired end state without going through the previous phases. However, this doesn’t mean that phases cannot be partly run in parallel.
In order to scale with AI, you first have to become data- driven.

According to a Harvard Business Review article (Bean and Davenport, 2019), “77% of executives report that business adoption of big data/AI initiatives is a major challenge.” Fewer than 10% cite technology as a challenge, and the article mentions various other general reasons why becoming data- and AI-driven remains a challenge.

THE BIG DATA AND AI JOURNEY

The data and AI journey plays a central role in this book because important decision making depends on the phase of the organization’s journey (i.e. its data and AI maturity). At a high level, the journey moves from exploring the opportunity to consolidating the data organization and benefits. The more an organization advances on the journey, the more data can be transformed into value in a scalable manner. In the later phases of the journey, full advantage can be taken of machine learning and AI techniques.

As an illustration I will discuss the data journey of Telefónica as depicted in Firgure 0.1. While each organization’s journey will be unique, the main phases, challenges and activities they work through will be more or less similar.

EXPLORATION

The first phase of the data and AI journey is one of exploration. This process can be started bottom-up, by data enthusiasts with technical capability (for instance, some data scientists or data engineers), or top-down, by managers who’ve heard about the benefits of data for business. Either way, an existing business problem (also known as a ‘use case’) is typically selected, such as reducing churn or increasing the effectiveness of marketing campaigns. Data is collected, and analytics is applied to see how it solves the problem.

At Telefónica, our exploration phase started in 2011 when we experimented with viral marketing. The objective of the use case was to increase the effectiveness of marketing campaigns for pay TV. Typical campaigns were based on individual customer profiles, and now we wanted to look into the potential of viral campaigns. We used call detail records (CDRs) of the fixed lines and extracted the embedded social graphs (who communicates with whom). Then, we identified groups that were characterized as having many communications, which we interpreted as representing social groups with strong ties. These were usually small groups of between four and six households. We then looked at groups where at least two households already had contracted the pay TV service. The hypothesis was that these two households would have a viral effect on the other members of the group, who would therefore have a higher propensity for buying the service than the average customer. The innovation was that, while customers had only been looked at in isolation, with this new technology their social relationships were also taken into account.

The results of this research were surprising. Looking at all customers included in the campaign, there was not really an effect, positive or negative. However, looking at the new customers resulting from the campaign, it turned out that many belonged to a group that was previously not considered as a segment: households with seniors (age 55 plus). Within this group there was a clear viral effect, and this allowed us to design a specific campaign that ended up driving a high conversion rate.

With these promising results, bringing together data, analytics and business, Telefónica started exploring ways to leverage data at a more global level, across its operations. A global business intelligence (BI) unit was created to get a consolidated view of what was happening in each of the operations, and formulate plans for sharing best practices and lessons learned. I had the honour of running Telefónica’s first global BI unit in 2012.

TRANSFORMATION

The objective of the next phase, transformation, was to prepare the organization to treat data as a strategic asset and create value from it in a systematic manner. Activities that were started in this phase included the selection and implementation of strategic use cases, which were deemed to have significant impact on the business. Use case selection happened at the global level, utilizing a kind of ‘menu,’ with possible options, but implementation was decided locally in each of the operating businesses.

In this phase, we also started a global big data road map to transform the traditional BI practice, which had relied on vendor data warehouses, into a more open big data architecture. At that time, Telefónica selected the Hadoop distribution of Horton works as the reference architecture.

Another activity we initiated was the creation of a data sourcing strategy. The telecom sector has a wealth of data available (CDRs, network data, web data, apps data, call centre data, etc.) but collecting and storing it all is not a trivial matter. This data often sits in vendor systems, and it isn’t always clear if and under what conditions telecom operators have access to it. Vendors became increasingly aware of the value of data, and in the early days we found that they weren’t always willing to provide access to it. An important lesson we learned for contracting with vendors was to always add a clause on data access. With these limitations in mind, Telefónica started to build a data collection road map, progressively adding more data and prioritizing it based on use cases.

Breaking down silos is one of the less technical challenges that comes with the transformation process. Data sources are always associated with certain business processes, and the business function owner traditionally had control over data generated by the business process. This data owner could decide how to extract key performance indicators (KPIs) from it, and with whom to share subsets of data. When it became clear that data held value, data also became a source of power. Sharing data with other areas of the company was sometimes perceived as losing control and power. With time, however, this resistance weakened, and now all line of business owners see the value of data sharing for the benefit of the organization.

One of the turning points in the transformation phase, and in the data journey itself, was the requirement to make budgeting for big data explicit in the annual strategic plan of all businesses. This made visible what until then was ‘hidden’ in IT and other budgets. For the first time, it became clear to everybody — from rank and file employees up through middle and senior management — how much each business invested in data and how much value it expected in return. There were many exceptions to the expectation that bigger businesses would invest and harvest more from data. And, this seemingly simple act of making the budget explicit, helped facilitate the requisite culture change. Until then, data professionals had to convince businesses to use data; now businesses were asking data professionals to help them.

When Telefónica started its data journey, the data ‘department’ was about six reporting steps away from the CEO. By the end of 2019, the Chief Data Officer (CDO) was reporting directly to the CEO. With increasing data maturity, the whole notion of data’s strategic importance has slowly progressed up the corporate ladder.

DATA-DRIVEN

Being in the data driven phase of the journey means that many of the important company decisions are now informed by data – that is, by data-rich insights, in addition to conventional wisdom, experience and intuition. There are, however, still important challenges to overcome. All companies in this phase have appointed a CDO, or similar executive, who heads a data team. But, what still needs to be achieved is the democratization of this capability, so the benefits aren’t only generated by a select group of data professionals, but by every employee in the company. Scaling up the value creation from data is more a cultural issue than a technological one. In this book, we will see how this process can happen.

Another activity that might start in this phase is that companies begin thinking about other ways of deriving value from data. So far, value has been mostly created internally, to improve the business. For some sectors, though, insights generated from first-party data can actually create significant value for other sectors. In Telefónica this is the case for the data that flows through the mobile network. Today, most people have one or two mobile phones, in addition to other devices, connected to the internet. All these devices generate activity in antennas, and this enables the generation of insights from anonymized and aggregated data gleaned from each antenna. Mobile network activity generates footfall and mobility data, which is of great value for sectors such as transportation, tourism, public administration, retail and finance.

Insights from mobile data not only have commercial value, but also social value. In 2016, we set up a dedicated department called Big Data for Social Good (BD4SG) in Telefónica that exploits insights from mobile network data — in combination with open data and other first party data — to help contribute to achieving the United Nations’ Sustainable Development Goals (SDGs). This effort involves close collaboration with humanitarian organizations and other NGOs working on problems such as forced migration, the human toll of natural disasters, child poverty, mobility’s impact on climate change and contagious diseases.

ARTIFICIAL INTELLIGENCE

The final phase is where full value can be created from the data, through analytics, machine learning and other AI technologies. This is also where companies might want to reconsider their original data strategy, based on several years of experience and learning, with the aim to scale even more. This may relate to revisiting technological decisions (such as on-premise versus cloud), organizational decisions, or relying on a centralized data team versus a distributed team. It may even extend to creating new business units, while also reconsidering local versus global responsibilities.

In Telefónica, this phase started with applying AI to change the way we interacted with our customers, using so-called ‘cognitive computing.’ This involved using Natural Language Processing (NLP) technology to automatically understand customer intentions, and then connect these directly to individual customer records to answer customer requests.

In this phase, we also started to view Telefónica as a platform company consisting of four layers. The first platform corresponded to the physical infrastructure (the network, antennas, fibre, shops, etc.). The second layer corresponded to the IT systems to operate the business (so-called OSS and BSS1 in the telecom industry). The third platform corresponded to the digital services running on top of the other two platforms. The key point is that these first three platforms generate huge amounts of data, and traditionally this data was kept (or not) locally. This is why it always takes so much effort to get decent data needed for use cases: it was dispersed across the company, technically in different physical systems, with different formats and vendors, and owned by different business users. And finally, the fourth platform is a new one that collects all data from the other platforms in an interoperable data format with clear semantics. This fourth platform is now the basis for all of the company’s data and AI initiatives.

Making maximum use of big data and AI also brings new risks, particularly related to privacy and the undesired consequences of AI and big data. Putting data and AI at the heart of an organization requires more attention to keep- ing the data of customers safe, for legal compliance but also to establish and maintain trust. Companies that use data, much of which is generated through how customers interact with company services, have many legal obligations in the European General Data Protection Regulation (GDPR), but also run the risk of reduced customer trust if the data isn’t used in a transparent manner. Also, the use of AI through- out the organization requires that companies make sure this technology is always used in a responsible manner. This means avoiding discrimination, putting humans at the centre, and opening black box algorithms when necessary.

Each organization will define or experience its own specific journey to become data-driven and AI-powered. As we said earlier, no two journeys will be alike, but there will be many common themes across them all. This book captures these commonalities such that each organization is able to take informed decisions on the key decision points throughout the journey.

ABOUT THE AUTHOR

RICHARD BENJAMINS is Chief AI & Data Strategist at Telefonica. He was named one of the 100 most influential people in data-driven business (DataIQ 100,2018). He is also co-founder and Vice President of the Spanish Observatory for Ethical and Social Impacts of AI (OdiselA). He was Group Chief Data Officer at AXA, and before that spent a decade in big data and analytics executive positions at Telefonica. He is an expert to the European Parliament’s AI Observatory (EPAIO), a frequent speaker at AI events, and strategic advisor to several start-ups. He was also a member of the European Commission’s B2G data-sharing Expert Group and founder of Telefonica’s Big Data for Social Good department. He holds a PhD in Cognitive Science, has published over 100 scientific articles, and is author of the (Spanish) book, The Myth of the Algorithm: Tales and Truths of Artificial Intelligence.

LinkedIn: https://www.linkedin.com/in/richard-benjamins/

THE BIG DATA AND AI JOURNEY

EXPLORATION

TRANSFORMATION

DATA-DRIVEN

ARTIFICIAL INTELLIGENCE

Suggested Reading