What does a Chief Data Scientist do?

What does a Chief Data Scientist do?
What does a Chief Data Scientist do?

Over the past decade, data scientists working in the corporate world have transitioned from being ancillary providers of value to becoming a key part of their firms’ corporate strategies. This evolution has occurred as part of a larger trend of digital transformation and includes efforts by businesses to centralise their data and to integrate analytics-driven insights into both their traditional and non-traditional product and service offerings.
As a result of the increased importance of data science, data scientists are attaining increasingly senior roles within their firms. The senior-most corporate data scientist is often referred to as the “Chief Data Scientist” (CDS) or “Chief Data Science Officer”. Combining a technical knowledge of data science with strong business acumen, the CDS is responsible for ensuring a business’s data science activities generate a successful return on investment (RoI). 

As the CDS role becomes more common, the skills necessary to perform it successfully have become more defined. This means for data scientists who aspire to move into the role, an understanding of the skills it requires is a necessity... a CDS is more than just an experienced data scientist.

What is a Chief Data Scientist?

The chief data scientist is the person who bears ultimate responsibility for the success and failure of their firm’s data science initiatives. They must act as the bridge between the technical and business sides of their firm, communicating its commercial needs to its data scientists and its technical needs to its chief executives. Their role requires a broad skill set that combines’ a data scientist’s normal technical proficiencies with a strong business acumen and communication skills. 

The chief data science role has become more prominent as many organisations have realised that they need someone who can provide their data science projects with strategic leadership. While the value of analytics is indisputable, three-quarters of companies aren’t seeing returns on their data science investments. Because a CDS understands data science from both a technical and a business standpoint, they are able to provide the guidance necessary to ensure their firm’s data science projects create real value. 

The day-to-day responsibilities of a CDS can vary significantly per-organisation; a CDS working for a small firm is more likely to have technical responsibilities than the CDS of a large corporation. The latter may frequently be involved in reviewing code and making technical decisions, whereas a CDS at a large firm will focus on administrative tasks, such as a review of key performance indicators (KPIs), project development, and resource allocation.

Role and responsibilities

The CDS must possess the ability to make executive decisions about how to complete data science projects successfully. Some examples of decisions that a CDS may need to make when administering the data science projects at their firm include: 

  • Deciding if their team should keep refining an existing analytics solution or invest time and resources into exploring whether new models could offer better performance. 
  • Assessing whether the time necessary to improve the performance of a project would cost more than the value that the additional performance would generate. 
  • Assessing whether their team has the expertise necessary to complete an upcoming project. If not, the CDS must determine whether any skills gaps can be filled by providing their staff with training or if it’s feasible to hire someone who holds the required skills. 
  • Ensuring that all insights created by a project can be understood by the non-data scientists who need to understand them, such as the CEO. 

Corporate data science projects are only useful if they generate more value than they cost; the above examples are all issues that a CDS may need to consider to ensure that their projects deliver strong RoI. To create value for their firm, the CDS must be able to understand how to identify opportunities for commercial gain and to work effectively with the C-Suite, while also possessing the technical skills necessary to oversee their projects to successful completion.

Decisions that a Chief Data Scientist makes

Case example - Improving Online Sales Conversions 

The following case provides an example of the decisions that a CDS at a large retailer might have to make in order to improve the revenue generated by their firm’s online sales portal. Throughout this process, the CDS must keep their superiors reasonably informed of their project’s progress in order to ensure that they have the executive buy-in necessary to ensure the project will be implemented successfully upon its completion. 

  1. To start, the CDS uses their business knowledge to determine what KPI to improve in order to maximise profit and then identifies a data science-driven solution to do so. For this example, the CDS selects sales conversions per-site visit to be their KPI. 
  2. Because they have a knowledge of the data science trends in their industry, the CDS knows that many retailers use predictive analytics to power their online storefronts. These predictive algorithms enable firms to provide customers with personalised product recommendations, which maximizes the likelihood they will make a purchase. 
  3. The CDS determines that a predictive analytics system is the best way to increase sales through their online storefront. This type of system is called a “recommendation engine”. Recommendation engines are a common form of predictive analytics used in consumer-focused machine learning applications, and power the music and film/TV recommendations used by services such as Spotify and Netflix. 
  4. The CDS must now oversee several key decisions, such as how many data scientists should be attached to the project, what language and tools should be used to build it, and how it should access and store customer data. These questions require the CDS to consider the effectiveness of each potential solution as well as its development costs.
  5. A project’s development costs are the money and time necessary to complete the project, plus the projected risk that something in the project could go wrong (further increasing costs and/or reducing the project’s end value). The CDS may know that graph databases are a high-performance solution for storing the information that their recommendation system relies upon, but they would also be aware that most data scientists do not have experience with graph databases, which could raise issues. 
  6. If the data science team lacks experience with graph analytics, the CDS must consider whether a graph-based recommendation engine is the best option to pursue. A graph solution may provide the best performance in theory; however, the increased development time and risk of errors that result from the team’s inexperience could ultimately result in it producing lower RoI than other potential solutions.
  7. In this context, the CDS must decide whether a non-graph based solution could provide sufficient RoI, or consider hiring an outside firm with graph database expertise to consult on the project. The “correct” decision depends on the particular needs, resources, and goals of the business—all factors that the CDS must be able to accurately consider as part of their role. 

A CDS may have considerable authority to make the above judgment calls by themselves, or they may need to consult extensively with their superiors. A CDS who reports directly to their firm’s CEO will generally have more authority than those who report to a lower-level executive such as the CIO. Whatever the case, the CDS retains ultimate responsibility to ensure that the projects under their purview are completed successfully.

The key skills needed to become a CDS

Most chief data scientists hold a graduate degree; however, it is less important to have a PhD than it is to possess the communication, business, and leadership skills that the CDS role demands. Every qualified data scientist possesses technical skills, but a far smaller portion also possess professional-level communication and business skills. This is largely because most data science courses focus on teaching technical competencies and pay little attention to teaching their students the business and communication skills that are vital to the CDS role.
The University of New South Wales’ online Master of Data Science program hones all three skill sets that a CDS needs. The program includes a significant focus on the use of data science for business cases, including dedicated courses on both strategic decision making and data visualisation/communication. Graduates from the program possess both the technical and administrative skills necessary to advance into—and succeed at—the chief data scientist role. 

UNSW’s data science program follows the university’s well-established reputation for preparing its students to succeed in the workforce—the school is one of Australia’s top 3 schools for graduate employability. Their data science curriculum focuses on teaching students the skills that are the most in-demand and the hardest for employers to find, maximising their competitiveness within the labour force. 

As corporate digital transformation initiatives continue to deepen the importance of data science to the business world, the chief data science position has become an increasingly important role. UNSW’s focus on teaching data science skills in a business-ready context provides the skills necessary for graduates to pursue that role. To learn more about UNSW’s Master of Data Science, get in touch with our enrolment team on 1300 974 990.