Processes for the Analytics Team


Once you have decided where the analytics team will reside within the organization, and have determined a method to fund its activities, it then becomes necessary to put processes into place for how the analytics team will function. Defining the processes should be a collaborative effort between the business and analytics teams and is one of the later-stage operational aspects of how organizations can get started with advanced analytics. Defining processes will not only result in better, more comprehensive and inclusive results, it will also give the business a stake in the game and help develop advocacy for the analytics program.


One of the most critical components of onboarding new analytical professionals into the organization is to build up their business and organization agility. This can be achieved through reviewing business process diagrams, shadowing various subject matter experts (SMEs) through their daily work, and discussing vision and goals with business stakeholders. It is also critical for new team members to understand what data are tracked in the organization, what data sources are available, and what level of data quality can be expected from those sources.

Promote Data Literacy

Data science professionals understand the connection between business processes and the data, as well as the meaning of the data in different contexts. Encouraging this literacy across the organization, both inside and outside the analytics teams, can be an important part of growth. Analytics leaders should strive to be a part of these initiatives.

Operationalize Solutions

As analytics solutions are developed, processes are needed to operationalize and manage these solutions. The analytics solutions will require validation and maintenance. There are plenty of standard methodologies for software versioning and testing that should be used for analytical tool development at the appropriate level. However, analytics solutions require an additional level of validation that is not commonly considered in general software development.

For analytics solutions, it is fundamental to ensure that the analytics models accurately reflect the key components of business processes as well as the business’ understanding of the problem. Analytics professionals should ensure the outputs of their models align with the business intuition of the corresponding SMEs and historical data. If the alignment does not exist at the level expected by the analytics consumers, then the model, data, or intuition should be refined until the expectation is met. This can be achieved by frequent reviews of detailed model solutions with the SMEs and detailed investigation of any discrepancies.

Institutionalize Collaboration

There are different ways that collaboration software can be used to enhance collaboration among data scientists, as well as collaboration with the analytics consumers and enablers. The most commonly used approach for selecting collaboration software is to use one of the technologies already available and supported within the company. Increasingly, big data analytics tools and software are incorporating capability for collaboration directly within their products. The particular choice of technology will depend on the team’s culture. Collaboration will help with data collection across the institution, which is a vital piece of your overall analytics strategy. Fundamentally, being able to collect more data theoretically enables an organization to utilize more business intelligence and analytics capabilities.

Establish Documentation Guidelines

Similar to versioning and testing, a set of standard documentation guidelines should be created and followed when developing analytics solutions. In addition, the key component of the analytics solution documentation should be the business process assumptions and simplifications, as well as assumptions about data quality and integrity. No analytics model will represent the precise details of the underlying process, so it is critical to reference not only what assumptions were made, but also why they were made, what solution design decisions were derived from these assumptions, and the impact of assumption violations on the solution quality. 

Up Next

Managing Analytics Talent