As a data scientist, I have the rare privilege of possessing the job title that Tom Davenport and others have dubbed the sexiest job in the 21st Century. As this popular job title catches on, I’ve even noticed a trend where customers make direct requests for help specifically from “the data scientist.”
Friends who don’t understand what I do often ask if I wear a lab coat to work. Of course, I don’t, but at times I have considered it! My typical day involves helping colleagues and customers find solutions to their most critical business needs through the power of analytics.
Though the title has gained rock star status in some circles, the truth is, data science and analytics cannot be successfully executed by data scientists alone. We are only one piece of the puzzle. Success in data science and analytics requires an entire team of employees in diverse roles. The data scientist would inevitably fail in solving critical business issues without a full team engaged in the analytics process.
What are those roles? And how can you build that dream team for analytics success? I’ll describe below the critical roles that you will need.
The business manager
The business manager role is someone on the data science team who understands how the business works. Sometimes referred to as a "subject matter expert" this should be someone with high degrees of business domain experience or business acumen. As the business areas change focus from one analytics effort to another, the business manager role should shift as well. The business manager for a customer-centric modeling effort would likely be different than the business manager needed for an employee- or workforce-specific effort.
For a data science team, the business manager is engaged very early in an analytics effort. They frequently assist in understanding the business need and translating that into a defined analytics effort. Often customers and business stakeholders can articulate their business need, but they don’t have enough analytical depth to articulate specific quantifiable questions. This is where the business manager can help! They can help take a high-level business need and formulate a specific problem or question to which analytics can be applied.
Also, after the analytics efforts have been completed, the business manager can help prepare and present the results. Because they have domain expertise, they can show the return on investment and help communicate the results in a way the business customer can understand.
The business analyst
The business analyst role on the data science team is to be the expert on the underlying business data and source systems. Depending on the background of the business analyst, it can be possible for the person fulfilling this role to also fulfill the business manager role as well.
Ultimately, the business analyst should have a good understanding of the fields in the source systems. As a result, they frequently assist with the data preparation and initial data exploration phases of the analytics effort. Their expertise is utilized to ensure the data management expert and statistical modeler understand the nuances, business definitions, and any needed transformations within the source data.
If the scope of an analytics effort is limited to data exploration and data visualization, depending on the technical background of the business analyst, this analytics task could be addressed entirely by the business analyst without the need for a statistical modeler. On the other hand, if advanced analytics models are required, the business analyst’s engagement is still required to help the statistical modeler validate the interim model results. The business analyst can help the statistical modeler identify possible focus areas or even reject approaches proposed by the statistical modeler based on their knowledge of the underlying data and source systems.
The data management expert
The data management expert is responsible for preparing and integrating the various source data and, when applicable, deploying the final analytics model. Typically this role would fall into an IT/systems management group. As mentioned above, in preparing the data, the data management expert will need to partner with the business analyst to ensure all business transformation and exceptions are addressed.
Furthermore, if the goal of the analytics effort is to create a repeatable analytics model that can be run on a regular basis, the data management expert would own the tasks associated with deploying the final model into a production environment (called model productionalization).
It's important to note that typically only a small portion of the data fields prepared and evaluated are used in the final model. Therefore, as part of the model productionalization, the data management expert will need to create a smaller productionalized version of the data with only the fields needed by the final model.
The statistical modeler
This role can be referred to by many names – data miner, statistician, data scientist, and analytical consultant. In essence, the statistical modeler is responsible for performing the advanced analytics methods that build a statistical model. If the goal of the analytics effort is merely to create a data visualization or perform some data exploration, as mentioned above, it’s possible a statistical modeler is not always required if the business analyst has the needed technical background. However, as the analytics initiatives begin to expand into the higher levels of analytics (such as data mining, predictive analytics, and optimization), I’d recommended engaging a statistical modeler.
The statistical modeler will work closely with both the business analyst and the data management expert. In preparing for an analytics effort, the statistical modeler will provide the requirements to the data management expert regarding how the data needs to be structured and prepared for the given analytics planned. Furthermore, when preparing to build the model, the statistical modeler will also work closely with the business analyst to understand the meaning and expectations for the source data provided. The interactions with these groups will continue throughout the analytics effort as the models are evaluated and even productionalized, as needed.
If the statistical modeler has an intimate knowledge of the business domain and the associated transactional data, it’s possible for the same person to fulfill both the statistical modeler role and the business analyst role.
Putting it all together
When working to build a team with these roles, the final makeup can vary. It’s possible to have multiple team members accomplishing a specific role. It’s also possible to have team members with diverse backgrounds take on simultaneous roles. Regardless, when preparing to tackle your next analytics initiative, taking the time to form a team with all the needed roles will ensure your project is on the right path for success.
Learn more about data scientists – and read profiles of other data scientists – in our new data scientist series.