Chief Data Officer Role & Challenges

Big Data has been replaced by Machine Learning and AI as the next “must have” trend. Machine learning has caught the attention of venture capitalists.

Chief Data Officers, Chief Analytics Officers,  Chief Data Science Officers and Chief Digital Officers are everywhere. The job is to leverage the latest in predictive analytics, data science, machine learning, and multi-tenant cloud architecture to bring innovation to traditional processes.

This is a pivotal moment in data driven business models (“systems that learn”) but there is no getting around the inherent difficulties associated with either altering organizational behavior, data ownership politics or managing transformation of the data infrastructure. And while the challenges are real, many firms are getting closer to achieving a data science and data management environment.

What are data and analytics officers overseeing… A variety of foundational & plumbing strategies:

• Data-as-a-Service:  Data Provisioning, Management, Lineage, Quality
• Reporting-as-a-Service:  Dashboards, KPIs, Drilldowns/Aggregates…. Descriptive
• Analytics-as-a-Service:  Predictive  Modeling and BI… Prescriptive analytics
• Information-as-a-Service:  Threshold based Alerts, Exceptions, Mobile Prompts
• Insights-as-a-Service:   ML/AI based “Systems that Learn”…automated learning – augmented intelligence, Next best Offer/Action

At core of all these, Data Management and Data Science tools are core technical and business capabilities.  Some firms are more mature and further along than others.

Why Mature Data Management as a Function

Organizations live or die by the quality of their data.  

Data is an underlying factor of input into business operations and essential in order to  facilitate process automation, digitize operations, support financial engineering and enhance customer facing analytical capabilities.

An effective data management program requires a planned strategic effort

• Integrate multi-discipline efforts
• Inculcate a shared vision and understanding
• Data is a ‘thing’ – vital infrastructure element foundation of the n-tier architecture
• Not a project, more than a program…it’s part of the core foundation

There is no question about it – the foundational levels of people, process, governance and technology required to establish data management on a  sustainable basis are coming together under the CDO umbrella.

What does a Chief Data Officer (CDO) do?

Capturing data is just the first step.

Most firms already have some semblence of data management in place for structured data.  Usually this is at the LoB level. The challenge is achieving an organization-wide perspective.

So how to do this? Every Chief Data Officer (CDO) has to do a mix of re-engineering processes, re-architecting and re-factoring to move the ball forward for structured & unstructured data.

The core elements of a CDO charter:

• Data Management Strategy: defines the framework for the data management program including the goals, objectives and scope; why it is important; how it will be organized, funded, governed and practically implemented
• Business Case & Funding Model: provides the justification for the data management program including the rationale for the investment; the costs, benefits, risks and expected•  outcomes; the mechanism used to ensure sufficient allocation of resources; and the  approach used to measure costs and contributions from implementation of the data  management program
• Establish the Data Management Program: identifies the organizational requirements needed to stand up a sustainable data management program including the operational framework to ensure sustainability and authority as well as the mechanisms to establish and confirm  stakeholder engagement related to program implementation
• Data Governance: defines the rules of engagement necessary for program implementation including the definition of policies, procedures and standards as the mechanisms for alignment among stakeholders
• Data Architecture: focuses on the core concepts of “data as meaning” and how data is defined, described and related; the identification of logical domains of data; identification of the underlying physical repositories; and the governance procedures necessary to  ensure the control and appropriate use of data
• Technology Architecture: addresses the relationship of data with the physical IT  infrastructure needed for operational deployment including how data is acquired, stored,  distributed and integrated across the organization
• Data Quality: establishes the concept of fit-for-purpose data; defines the processes associated with establishing data control; and addresses the implementation of governance mechanisms for management of the data manufacturing chain of supply
• Control Environment: defines the data lifecycle process and how data management is integrated into the overall organizational “ecosystem”

In each initiative above,  CDO have to conduct a “current state” to “future state” transformation assessment.. skillset, mindset, dataset and toolset also have to be evaluated.

Figure 1: Enterprise Data Management

Making it Real: Chief Data Officer Challenges

Chief Data executives have been hired in many companies and given the authority they need to drive data management and data science initiatives forward.

Many firms is now standing at the precipice of turning that “data management commitment”  into “the art of the  possible” action. Many firms are in the midst of re-writing internal policies and standards needed to embed data management into  the fabric of their organizational operations.

Challenge 1: Foundational levels of governance are in the process of  being established

• Data owners (i.e. chief data officers, or CDOs) have been hired and tasked with addressing the gaps and challenges associated with data management
• The “Office of Data Management” has become an official control function with defined processes and has been provided with both the executive air cover and authority required to integrate data management into the organizational environment
• Seed funding is in place to get data management programs underway
• The political challenge is to scale these pilot initiatives across the organization

Challenge 2: Changing organizational behavior is difficult

• Data policies and standards are in the midst of being created but are undergoing rigorous scrutiny because the adoption of policy mandates compliance (and many firms would not be in compliance with their adopted policies)
• Data stewardship and accountability are defined and in the process of being integrated into the operational processes of the organization
• Business buy-in is still tentative because many firms are benefiting from seed funding to get their data management programs underway

Challenge 3: Implementation of the data management infrastructure remains a priority

• Critical data elements  (CDEs) and critical data attributes have been defined but not fully inventoried or aligned with compounding processes
• Unraveling lineage, mapping complex data flows, separating data attributes from calculation processes and building data inventories are huge tasks that require time and cross functional collaboration
• Adopting unique identifiers and harmonization of content to precise contractual meaning across hundreds of repositories is in process but remains a daunting challenge

Challenge 4: Data quality control procedures needed to ensure trust  in data remain elusive

• Profiling and current-state assessments needed to prioritize remediation are not very advanced
• Data quality control processes including the adoption of business rules and the establishment of authorization points are still in the early conceptual stages of development
• The industry is still relying on tactical “find” and “fix” approaches to data reconciliation

Summary

Creating a common terminology;  Shared understanding among stakeholders; A clear path to increasing capability maturity — are all things CDOs have to do.

The good news  is that leading firms are overcoming the organizational barriers required to enable data driven business models  as a sustainable activity.

The more challenging news stems from the fact that the scope of the task is significant. Large enterprises are complex  organizations who are forced to deal with the intricacies of multiple LoBs, the need to unwind technical legacy and the  pressures of volatile global business environments.

Most firms are working to untie the “Gordian knot” and dedicating  resources to data science models, new platforms like Spark, unraveling lineage, inventorying content, identifying critical data, adopting standards and mapping systems. Harmonization of meaning across thousands of repositories and implementing control processes needed to ensure trust in  data resources remain as daunting challenges.

Additional References:

1. Data Management Maturity (DMM)
2. EDM Council