News

Critical factors to consider when assembling a data analytics staff

Posted

A McKinsey Global Institute study estimates that by 2018, there will be a shortage of 140,000 to 190,000 people with the necessary analytical expertise, and a shortage of another 1.5 million managers and analysts with the skills to interpret and make decisions based on analytics (James Manyika, 2011). With such a skilled labor shortage, and an explosion of data from networked devices, there are vast opportunities for analytical research and development. Consequently, publishers with analytics needs should have a strategy for developing an in-house analytics staff. The most important ingredients of this strategy include identifying the data roles that the publisher needs, a culture that is data-driven and customer-centric and a repeatable plan for managing and developing in-house analytics talent.

Publishers with data needs often focus too much time on their data and not enough time on the data roles they need to fill and the hiring of people with customer-centric skills (Matt Ariker et al., 2013). Typically, a combination of project manager, data scientist and data engineer are necessary to build an analytics staff. There will be considerable overlap in these roles in smaller media companies, and an opportunity to grow a staff one person at a time, scaling the team as desirable results dictate. The project manager should be experienced in managing quantitative projects, be able to roll up his sleeves and code if necessary and understand quantitative algorithms. The data scientist will have a quantitative background as well, plus research experience and solid communication skills. Unlike in the typical software development world, there’s not as much focus on very specific technical skills. Candidates that has done formal research or that has solid statistical skills are a good start. Often, though, the data scientist will have a computer science or mathematics background. Lastly, the data engineer must understand software engineering fundamentals, along with computer science topics such as data structures and algorithms (Rivera, 2015).

Along with technical and problem-solving skills, data science team members need a customer service first mentality. You can hire a technically proficient staff that builds sound products, but if value is not added, then failure is imminent. For example, if the end result of a campaign to program content to match the audience engagement results in low click through rates or conversion then the team has failed.  Ongoing communication inside the publishing company and methodical experiments by the analytics team will mitigate this risk. The analytics team must be able to communicate in a language that the publisher understands. This involves telling a story through analytics and verbal communication that meet the publisher’s needs and add value.

Analytics projects are not dissimilar from conventional data warehouse projects in that success is directly proportional to meeting the publisher’s needs. Machine learning and statistical models are not “one size fits all.” There will be decisions to make regarding quality, timing and cost. These metrics must be put in context regarding the publisher’s project needs. This all underscores that a major component of an analytics staff member’s role is to educate the media company about what is possible with analytics and the impact of lower quality, tighter timelines and lower budgets that result from the options available (Roger Stein, 2015). An analytics staffer with the proper balance of technical and communication skills is uniquely qualified to foster the data-driven and customer-centric culture necessary to solve problems effectively.

A data-driven and customer-centric culture is key to empowering effective analytics teams. An example of an organization that is data-driven is LinkedIn. In the early days at LinkedIn, analytics staffers were empowered to experiment and get real feedback from analytics. One such experiment evolved into ubiquitous functionality at LinkedIn, and a huge financial success. A LinkedIn scientist got the idea to test what happens if you show users the names of other users that they are likely to know and to connect with. The scientist, Jonathan Goldman, created a custom ad that identified the three best connection matches for a user based on his profile. After a few days, Goldman’s ad had the highest click-through rate LinkedIn had ever seen. His “People You May Know” ad had a 30 percent higher click-through rate than other LinkedIn ads, and was the catalyst for millions of new page views (Davenport and Patil, 2012). 

The data-driven and customer-centric culture lays the foundation for the analytics talent management strategy. The shortage of skilled data and analytics workers makes it imperative to have an analytics talent strategy. A The McKinsey study found that 15 percent of operating profit increases from analytics projects was linked to the hiring of data and analytics talent (Bughin, 2016).  With this talent in place, analytics can be used to address a publisher’s business challenges and to develop new applications. Since it can be cost prohibitive for a publisher to staff solely from established data scientists, an alternative to recruiting a whole staff of seasoned scientists can be to groom analytics talent from within. The existing business knowledge in a media company is extremely valuable, and an employee that adds analytical skills to his existing business knowledge can bridge the gap between purely technical and purely business employees. Also, with so many media companies competing for analytics talent, it makes sense to look internally to fulfill analytical staffing needs.

When building an analytics staff, it is important to be conscious of hiring employees that have the skills that align with your publishing needs. Data science curriculums in universities are becoming popular, but, unlike engineering schools, there’s not a standard curriculum. Additionally, the shortage of analytics talent necessitates a talent management strategy for any publisher that plans for analytics as a core competency. This strategy should encompass a plan for hiring external staff, training or growing internal staff and cultivating a culture conducive to analytics and data-based decision-making. This culture should provide employees the freedom to experiment endlessly and to explore all possibilities. Being customer-centric should also empower employees to communicate costs and trade-offs of the various inputs that go into analytical models, so that the publisher can always make the most informed decision.