There are organizational and technology components critical for a business to succeed in becoming data-driven. On the organizational side, a key component to succeeding with data and analytics is to create a culture that supports these efforts. Companies that succeed are typically goal-driven, transparent, empowering, and collaborative. They have strong leadership that believes in data and they are governance oriented. On the technology side, the company takes steps to ensure sound data quality and has operationalized analytics to take action. Data-driven organizations often have an integrated analytics and data management strategy that spans the entire analytics life cycle from problem identification to data access and manipulation through analytics development, deployment, and monitoring.
It can be difficult to create an organization that thrives on data and analytics. It involves breaking down silos and getting people to see eye to eye, as well as building trust and collaborative team structures. It also involves enabling more people to access and work with data, which requires both skills and tools that support different personas such as data engineers, DevOps, data scientists, business analysts, and business users. In my latest TDWI Checklist Report, I outline 5 best practices for becoming more data driven:
- Build relationships to support collaboration. A key is for groups to appreciate each other’s perspective in order to move forward effectively. Of course, having executive support can help because executives set the tone and vision for the organization. However, it is important for other data and business leaders to come together as well. That means sitting down and communicating. A typical comment we hear from organizations looking to become data-driven and collaborative is, “Once IT understood what we were trying to do, it became easier to work together.” Clearly defined roles and responsibilities are also important.
- Make data accessible and trustworthy. At TDWI, we are seeing organizations integrate and access data across a multiplatform environment for analytics. This is becoming the norm, as organizations need to analyze “new” forms of data (e.g., text, sensor, image, streaming) as well as traditional forms. Organizations are making use of platforms including the warehouse, Hadoop, streaming platforms, and data lakes, both on premises and in the cloud, as part of the move to a multiplatform architecture. Therefore, the key is to unify the data for analysis.
- Provide tooling to help business people work with data. Vendors have been working hard over the past few years to provide easy-to-use, self-service tools to help business teams with such tasks as data preparation and analysis. Many of these tools have advanced analytics “smarts” (such as machine learning) built into them and are designed to work across the analytics life cycle, from data collection and profiling to monitoring advanced analytics models in production. Important principles here include automation, explainability, and reuse.
- Consider a cohesive platform that supports collaboration and analytics. TDWI Research indicates that an analytics platform is a top priority for organizations looking to become more analytically sophisticated and data-driven. Because there are numerous roles that contribute to the overall analytics effort, vendors are beginning to create platforms that support multiple personas. For instance, it would not make sense to purchase a data science workbench solution that primarily supports those who code in Python if the average user is a business analyst. Business analysts might like a visual user interface where they can construct workflows.
- Utilize modern governance technologies and practices. TDWI research indicates that most organizations do not do a good job of governing their data. In fact, about a third of organizations don’t govern their data, at all. Governance is an evolving practice. As more people become involved in analysis and as organizations deal with big data, the cloud, and other new technologies (such as stream mining), practices need to evolve. For instance, as more people analyze data, there is a greater need for consistent vocabularies and flexible access to data. Similarly, as new kinds of data come into the organization, an enterprise must determine how to deal with them. It is equally important for organizations to have policies for implementing predictive models.
Ready my checklist report to learn more!