It’s easy to say that the demand for data analysts is skyrocketing. Back in 2017 IBM claimed that by 2020 the number of data science jobs that need to be filled will grow to over 2 million.
With so much demand, picking an analyst who can hit the ground running must be easy pickings, right? Well, not quite.
The reality is that the probably of a finding a professional who will take off immediately is not likely. The expectation of what professionals with technological knowledge should bring to the interview table has also skyrocketed. As a result managers entertain false assumptions about what skills candidates are realistically likely to offer. Instead they look for unicorns – people whose resumes demonstrate all the technological knowledge required by the organization. Such searches are bad milestones on a roadmap to building an developer team.
To select the right analytics skills, managers must consider what roles really need in-depth skills, and how to share training across the team to deliver up-to-date information. Good analysis definitely requires domain knowledge, but it also requires understanding if the right trade-offs are present when interpreting data and drawing conclusions.
Organizations should look beyond fixed skill-sets, and establish an analytics culture emphasizing problem solving over narrowly defined domain knowledge.
What this means in practice
For example, creating database queries is essential for building data models. Building the right query does require some technical skill. But the intuition for building the model means appreciating what kind of data is being requested. It may mean some familiarity with products and services to know which observations in the dataset can potentially solve a problem for customer needs, or for operational concerns within the business. Applying that intuition also means having the sensibility to ask the right questions to keep a project on track, and to avoid ad-hoc queries and questions that consume time and resources needlessly.
Fluid skill management is a good complement for an organization relying on CRISP-DM (Cross-Industry Standard Process for Data Mining). The cyclic nature of the CRISP-DM process is good for revealing how well those team can monitor and collaborate (I explained the value of CRISP-DM in an earlier post on dashboards for machine learning). An organization needs professionals with a consistent desire to learn, as well as a system that allows the learning to take place. An analyst may not have all the skills you want, but that same analyst, with the right learning capabilities, can help the team build the best queries, analytical models, and dashboards imaginable.
The value of diversity
Another benefit from emphasizing roles as much as skills is that it can help attract diverse talent. Not only does diversity provide a broader perspective on data and its sources, but it should also add more capability in addressing social issues that are impacted by analytic and machine learning initiatives. Since data often represents a social or physical reality in meaningful ways, developing a diverse team is an imperative for detecting and addressing social impacts from analytics and machine learning. This can be more important than an overfocus on technical solutions.
Developer quality does not necessarily require a deep understanding of code. When hiring smart people for analytics, you really want that person’s curiosity and experiences to drive the team capability. Their experiences become the glue that really keeps data diagnostics and delivery processes in step with organizational needs, and can even lead organizational progress.
That progress is essential for delivering products and services that provide customers with meaningful solutions.