7 Questions to Ask before Hiring a Data Scientist
Does your organization need the special skills of a data scientist? Before you begin a talent search, consider answering these seven questions first.
By Yannick Koger
"Data scientist" has become one of the hottest new job titles. Indeed, it's even been called
"the sexiest job of the 21st century." That's why more companies are planning to hire one or more. In some cases, hiring is driven by a mandate from upper management, who believe data scientists can help the company get a handle on big data. In other situations, an organization has a clear need to develop advanced analytic capabilities and generate more value from new and existing data sets. ()
To ensure they get the right scientist for the job, companies must think through several critical questions, including:
1. What's the business case for hiring a data scientist?
The answer will vary tremendously by company. From a business perspective, customer experience improvement can be an important pillar of increasing customer lifetime value (LTV). For example, using detailed data to identify common issues or bottlenecks in checkout processes or to improve recommendation algorithms can greatly simplify the path to purchase (which has obvious bottom-line implications). Improvements in targeting promotional offers and more sophisticated segmentation can improve marketing ROI. Efficiency gains for key operational processes can result from better understanding of sensor and machine data. Any area where data and analysis can be used to improve performance can potentially be included in the business case for hiring a data scientist.
2. What is a data scientist anyway?
The conventional definition holds that data scientists have a background in advanced analytics (e.g., statistics, mathematics, predictive analytics, and computer science) and/or data engineering/information management (e.g., data warehousing, IT infrastructure, cloud-based systems, HDFS, and MapReduce). Some experience as a business analyst and data mining and visualization work is also common.
That's all fine, but the purpose of a data scientist is just as important as specific skills or areas of expertise. A data scientist should also be able to shape a strategic vision for harnessing the power of data to transform business performance. He or she will spend considerable time discovering insights and then communicating those insights by telling a story with the data. Being a skilled communicator can also help navigate the likely cultural changes that will take place (see Question 1). In fact, communication is so important that "data evangelist" may be another sexy title for this position in the near future.
3. Is our organization ready for a data scientist?
Readiness indicators typically include:
- Data is increasingly viewed and treated as a valuable asset that has to be actively managed and understood
- Groups within the organization are looking for insights beyond reporting what happened; they want to better understand why it happened and -- more important -- what is likely to happen next
- Management is looking for leadership of its data and analytics-enabled capabilities -- perhaps a chief data officer or chief analytics officer has been hired
- A rapid increase in planning for projects related to Hadoop, unstructured data, big data cloud solutions, advanced analytical environments, and/or advanced business intelligence (BI) solutions
- Analytical horsepower is bogged down in basic reporting, and there is limited ability and/or no plans to figure out how to achieve "Aha!" insights that can lead to breakthrough performance gains or identify new revenue streams
All these symptoms may show a tactical need for a data scientist, but it's also worth asking if the company is culturally ready. The hiring of a data scientist may mark an important milestone in the adoption of a data-driven and analytics-enabled management style rather than decision making by gut instinct or executive seniority. As the business landscape evolves more towards fact-based decision making, a good data scientist can be the catalyst your company needs to drive cultural change.
4. Do we need one data scientist or several?
Many people have no clue what a data scientist is (even some of the people staking a claim to the title). That's partially true thanks to potentially vast gaps in capability. It's unlikely that any individual will be an expert in everything a data scientist could do. That's why some companies take a team-based approach and set up small data science exploratory units. By building strong teams at the outset of a data science program, companies can lay the foundation to evolve their capabilities and eventually develop cross-functional analytic centers of excellence. In other words, the most effective data scientist may actually be a team.
It's worth adding that, because there is a dearth of data science talent, companies must pay a premium when they find qualified workers. Today, only a select few individuals possess deep expertise in all or most of the "required" skills. The combination of the talent gap and high cost may mean a "fractional" data scientist resource from the outside may be a better fit. The expertise and objectivity of a part-time or consulting resource can be an integrated into a broader team.
5. Where should data scientists fit into the organization?
Given the enterprisewide impact of analytics and cross-functional nature of data flows, the organizational chart positioning may be less important than the scope of involvement and the freedom to explore data freely across the organization. There's little doubt that skilled data scientists can help improve performance at every level of the organization and across functions (from marketing to supply chain). Much of the value comes from comparing, cross-referencing, and exploring diverse data sets. Because scientists love to experiment, an environment that affords time for creative discovery is critical to long-term success.
6. What does a data scientist cost?
Because demand is high and supply is low (qualified candidates are sometimes referred to as "unicorns"), data scientists can seem expensive. Current estimates range from $100,000 to $125,000 per year. A few years of experience can yield salaries two or three times higher, according to one report. Some companies are turning to academic researchers to fill the gaps.
Of course, money will not be the only variable in attracting a data scientist. The current state of infrastructure and available toolsets are likely to be important considerations. Cultural factors may also come into play. Scientists love experimentation and finding answers, so their job description should reflect those characteristics.
7. Does a scientist need a "lab" or require a new set of tools?
The answer depends on the current state of the IT infrastructure, whether or not a comprehensive data strategy already exists, and the maturity of master data management and data governance programs. Assuming the data scientist's top priority is to establish an environment where high-quality data is assured, then it's more a matter of which analytics tools and solutions align with top business priorities. For instance, if improving customer intelligence is an important objective, then a data management platform (DMP) or CRM system can help the company achieve 360-degree customer views and provide a wealth of insights for data scientists. On the statistics side, a data scientist may have a strong preference among the many available options, including SPSS, SAS, R, Matlab, and Python, to name only a few.
Finding detailed answers to these questions can help your company get more specific and concrete about what a data scientist can and should do for your organization. In other words, taking a more scientific approach to hiring a data scientist is definitely the way to go.
Yannick Koger is principal consultant with Infinitive's big data and customer intelligence practice. His combination of skills and expertise in customer intelligence, data warehousing, advanced analytics, marketing strategy and data visualization techniques help him bridge the gap between IT, analytics, and business-unit performance. You can contact the author at