The chief data officer (CDO) is beginning to be recognized as the lynchpin for tackling one of the most important problems in enterprises today – leading the transformation to a data-driven culture. Often with a budget of less than $10M, one of the biggest challenges and opportunities for CDOs is making the self-service opportunity a reality by bringing corporate data assets closer to line-of-business users. Bigger budgets are on the horizon for CDOs who emphasize agile platforms and methodologies that allow resources, skills and functionality to shift quickly between CoE and LOB. Those budgets and the work they fund are growing faster than the talent pool, so CDOs will continue to face challenges as they compete for skilled data scientists.
Growth of Data Science vs. the Trained Data Science Pool
One of the biggest obstacles CDOs currently face is a lack of skilled talent for data science. IBM predicts that the demand for data-driven decision makers, such as data-enabled marketing managers, will comprise one-third of the data savvy professional job market, with a projected increase of 110,000 positions by 2020. Yet there remain very few graduates with enough data science skills joining the job market to fill these positions. While companies are pouring money into attracting and retaining talent, the most efficient way for managers to address this issue is to add technical skills and capabilities to their existing teams. It is easier to add technical skills than it is to add domain knowledge. Rather than spending time and resources competing with outside companies for talent and then spending more time ramping up new hires until they fully understand your company, spend that time upskilling the team you have in place. Many companies have identified this as the most practical (and cost-effective) solution—leading to a new breed of “citizen” data scientists. Citizen data scientists have the capability to understand the techniques of data science and create models that drive business value without an advanced degree in data science. Giving them the ability to start the data modeling process provides them the framework to talk to the data science team, which ultimately fosters collaboration that leads to higher levels of success across the organization.
Communicating the Problem and Results
There is a significant gap in the process for data scientists in communicating the problem and the results with the lines of business. While a nanodegree cannot replace a traditional master’s degree in data science, it can certainly educate citizen data scientists to a level that will speed up the feedback loop in communicating problems and results. Data scientists have fundamental, deep technical knowledge, whereas business teams typically have a greater depth of knowledge about the organization as a whole. This means that citizen data scientists that are coming from the business side of the organization can look at data and understand why a correlation exists between two columns from their own knowledge of the company, without needing to compute or see that written out as data scientists typically would. The data scientist is able to provide the factual relationship, while the business team can understand what the impact is, and what to do differently. This is one of the greatest benefits in creating citizen data scientists from within the business, as they can not only create models, but they can then quickly connect the dots between the results and the causality—ultimately leading to faster, data-informed business decisions.
Access and Correct Modeling
With the growth of data science, organizations are expecting more aspects of their company to drive value from data, which ultimately means that data scientists are spending too much time getting access to data and manipulating it into a form that can be modeled correctly across more and more parts of the business. While trying to foster a data-driven culture, data scientists run the risk of spending too much time in this low-value part of the modeling process. However, if you enable citizen data scientists from within your own organization, you can facilitate an overall higher level of productivity across the business. This ultimately comes down to effective management of resources and a CDO that can efficiently relegate assignments. Utilizing high productivity self-service tools and enabling workers at all levels throughout the organization is only one part of the puzzle here. An effective CDO should be able to assign the upper level data scientists to create models and solve problems that are essential to the organization’s bottom line, and then direct them to advise on lower value models that others are working on across the company. Given how much is being asked of CDOs, it remains essential that they make the most of the resources available to them, implement the most productive technology, and delegate data modeling approaches effectively.