The Role of a CDO. Do we need it?

Updated: Feb 25


Excuse me, AI?

It is not my favourite way to start a blog entry but, in nowadays, AI seems to be like teenage sex: everyone is talking about it but only a few are experimenting with it. My experience in the market is that most organisations have discussed, or even considered AI at some point, but only a few have started implementing internal "AI projects", and almost all of them are getting modest results compared to their expectations, which certainly doesn't facilitate future investments.

Enthusiasm about the potential of AI was, in many cases, unrealistically high. This was partially due to overconfidence in organisational capabilities, which were too immature to exploit, execute and deliver on that potential, as well as a profound misunderstanding of what AI is. Relegating AI to the technical domain and disconnecting it from business problems it could potentially help to solve or, at least mitigate, is a commonly repeated mistake.

Despite almost all senior managers being able to mention, with a little curiosity and skepticism, cases like Google Deep Mind, Amazon Alexa, or Apple Siri, very few (there are always exceptions, such as online retailers like Zalando and Zara implementing successful business-oriented usage of their data) seem to have any idea how to translate that theoretical potential into industry-specific success stories, and almost none can say how AI made their competitors better or faster or more efficient, even though almost all CTOs I spoke with seemed assured that AI would have wide-ranging effects within the next five to ten years.


At times, this super innovative industry tends to forget about the basics. AI might be the new electricity as Andrew Ng from Stanford stated (I personally agree...), but it just describes the "how" and not the "what". It is hard to imagine that an electrician can fix a broken fridge just because it uses electricity. For any data scientist, it would be difficult to solve a vertical problem without the appropriate domain knowledge. For any domain expert it would be difficult to realise what AI can do to solve a problem without a data scientist. The tension and miscommunication (sometimes they seem to use different languages) between those roles might explain why companies seriously struggle to get a return from AI investment (to their investors). Clearly, there is an ability gap in understanding both domains, orchestrating and leading data initiatives.

General Scepticism and uncertainty perception.

Concerning data science expertise and suitable products, there are some significant differences in levels of practical organisational readiness. Perhaps it is human nature to see barriers rather than opportunities, but it is also clear that there is a significant level of uncertainty and resistance by some organisations when it comes to embracing AI.

From an organisational perspective, the lack of trust in AI and the lack of the critical skills to drive from potential to the actions, are severe challenges to its uptake. To develop the trust needed both internally and externally, it is crucial for a data science team to develop a deep understanding of business issues and abandon the mantra of “data is just data, no matter the problem”, as well as executives being prepared to trust insights from the data. This journey usually requires a corporate culture change and would, therefore, take time. This reality explains why many senior managers are more optimistic about the potential of AI than their organisation’s readiness to exploit it.


Organisational and Cultural Challenges. The need for a robust Chief Data Officer.

Despite the skepticism and uncertainty, a good sign of the importance that leading corporations give to data is the establishment of the Chief Data Officer role. Recognition of the CDO role is confirmed by results of an annual survey of Fortune 1000 c-executives launched by advisory firm New Vantage Partners. The study was conducted in 2018 and the recently released report indicates almost two-thirds of executives reported their company has a CDO. This figure is remarkable when compared to 2012, when only 12% of the executives reported having a CDO. This indicates that the Chief Data Officer role has become a permanent role within most leading corporations and for good reason. Despite the trend of recognising the need for a Chief Data Officer, there is a lack of consensus on the nature and responsibilities of the role, as well as mandate, and background that qualifies an executive to operate as a successful CDO. Furthermore, because few organizations have assigned revenue responsibility to their Chief Data Officers, for most companies, the CDO role functions primarily as an influencer, not a revenue generator, and that might also explain the confusion and absence of unified vision about exploiting the potential of data and AI. The different view on CDO responsibilities, mandate, and importance underscores why Chief Data Officer is a challenging job in the c-suite within many organizations, and why the position has become a hot seat with a high turnover in several enterprises. Too often, the Chief Data Officer is not perceived as the executive with primary responsibility for data strategy and results within the organisation. Other C-executives, such as the CTO or CIO are considered the focal point. That illustrates some of the challenges that the CDO faces within many organizations. To increase the confusion, it is typical for C-level and VP level executives to openly admit that data responsibility remains in silos where each product or business unit keeps the full responsibility of the data and analytics initiatives. In other words, there is no single point of accountability for data and analytics within and across their organisation. For those professionals with more than 20 years’ experience, this should not be surprising. Most corporations that have been in existence for decades, or longer, were not organised around data as an organising principle, or as an enterprise asset. As companies adjust (too often requiring more than simple adjustments) to become data-driven organizations and work to forge data cultures, there is bound to be lingering resistance to change, or a profound lack of agreement on where data responsibility lies, or even what it looks like.

A CDO Role with a clear mandate.

Like any other new role in an evolving industry, the mandate of the CDO has different interpretations, depending on the vertical, the company size and organisations rigidity. There is a consensus that the CDO should be playing a leadership role when it comes to projecting the course of a company’s data and analytics strategy. However, there is some level of discrepancy when defining the primary responsibility for a CDO. While some companies see the CDO as the person to lead and develop overall data and analytics strategy, which implies a strategic role, other companies assign the primary

responsibility to coordinate data initiatives across the company, relegating the CDO to an internal highly paid consultant without the ultimate accountability over data projects. This second approach is typical of companies that, even making a genuine effort to make their products and offering “data- driven” or “autonomously intelligent”, have an unclear view of how to do it. Unfortunately, the big wave around Data Science and AI makes many companies articulate an external message without facing the required internal challenges and changes. In my experience, it is highly unlikely that those companies generate new revenue streams, either from their internal data study or by including AI features in their product portfolio. The CDO role adjustment is expected to continue and play out as companies progress on their data journey. Organizations need to remember that AI and Data-Driven culture is a process, a long one depending on the starting point and the resistance to change. It is merely premature for some organisations to assume the need of a CDO and data strategy.

What skills are required?

If something comes along with the change, that is uncertainty and divergences about how to build the future. Thatisalsotruewhenitcomestothebackgroundandqualificationsthatmakeforasuccessful Chief Data Officer. For those who decide to establish the CDO role, two opposite and almost equally adopted approaches are predominant. One school of thought is that the CDO should be an external change agent (outsider) who brings fresh perspectives. Almost equally accepted, some companies believe that the ideal CDO should be an internal company veteran (insider). Someone who understands the culture and history of the company and knows how to get things done within that organisation. As such, there is no correct answer to this question, but the level of conservativeness within an organisation, particularly in the top management, the aversion to risk and the resistance to change will mark the path to follow. The heritage, age, and dynamics of the company will also play a crucial role in the decision. Companies like Google, Facebook or Amazon, and many others, were born in the data age. Data is in their DNA. Most likely those companies do not need a fresh perspective because in many cases, they build the new industry views when it comes to monetising the value of their data. It is important to highlight that, despite the differences in the mandated scope and the CDO background (outsider vs. insider), there are a few key ingredients that everyone seems to agree with when it comes to defining a successful CDO:

  • Must be either a data scientist or a technologist who is highly familiar with Data Science practices.

  • Must have a deep understanding of the vertical where the company is operating.

  • Must be able to translate business objectives into Data Science activities and, Data Science insights into business actions to reach the target.


Financial accountability is, clearly, the major friction point with a profoundly different matter of opinion. Another controversial aspect is related to the seniority of the CDO, while some companies include the CDO as a member of the executive committee, other companies prefer the CDO to be reporting to the executive committee, without a seat.

Looking to the Future

AI and Data Analytics adoption for any enterprise of any size and business is a journey, and that journey requires a driver. This driver should oversee making the process safe, secure and profitable for the company. Like any other driver, experience is the critical factor. While the role of CDO has emerged as a de facto industry standard and gained broad acceptance as evidenced by the sharp rise in CDO appointments, agreement on responsibilities, mandate, profile, and seniority continue to vary substantially depending on factors like company age, heritage and business. The lack of consensus on the Chief Data Officer role illustrates the diversity of opinion on the value and importance of data as an enterprise asset and how it should be managed.

In my opinion, it is only a matter of time before this discrepancy is put behind us and data ready companies will be those able to compete faster, agiler and more efficiently. The CDO will be near the center of action as companies strive to become data-driven, but should also be prepared to face a hard transition towards the coming global data age.


About the Author

Roman Ferrando has numerous years of experience within some of the world’s leading technical companies like Ericsson Research and Oracle Labs. He has focused its research activity on clustering and pattern recognition techniques and the scalability issues related to hight-peed generated data. He holds 15 patents related to the analysis of real-time data in telecommunications networks and have also produced several academic papers on the same topic published in well-respected journals. He has been also member of the 3GPP standard committee for several years. Roman holds a PhD in Machine Learning and also authored a book on the role of data analytics in the cloud and next generation telecom networks. Wiley & Sons published this during 2013. In 2016, Roman started Thingbook, a company created to facilitate the adoption of predictive analytics in high-speed data generation environments.

0 views0 comments

Recent Posts

See All

After a few years of fantasy and unrealistic expectations, artificial intelligence (AI) has made the leap to practical reality ...

Finding and analysing anomalies in data purely using data observation without previously characterised knowledge (unsupervised learning), provides the Scientific Community with results that are both f