Meenal Iyer is data & analytics strategist and transformational leader with over 20+ years of experience building data platforms and driving enterprises to be data driven. Her specialties include Data Monetization, Data Democratization, Enterprise Data & Analytics Strategy Roadmaps, Data Science/Machine Learning, Enterprise Data Management/Governance
What does an analytics-driven culture mean?
It means that all decisions across the enterprise, any business decision making, and product innovations/strategies are driven by data and analytics.
First and foremost, an enterprise data/analytics strategy is key for any organization.
This strategy will contain a mission statement, a vision, and an execution plan of how to get there. The strategy must have the necessary executive (C-level and business stakeholders) support with the knowledge that the execution of this strategy is a multi-year initiative.
The strategy will contain a plan regarding data access either as a single source of truth or data as a single version of the truth.
In a single source of truth organization, all data within the organization is centralized within a single data repository. The data contained within the repository should be capable of providing data both in near real-time and batch modes. The onus of the data quality and ownership of the data resides with the Data Engineering teams.
In a single version of truth organization, common data for the enterprise will lie within a single data repository or warehouse, but a lot of the peripheral data will lie within the individual business units/data sources that will provide their data reporting capabilities. Here near-real-time and/or batch modes will be made available. The onus of the data quality and ownership lies within the data resident organization and is more decentralized. However, the system housing the data is the source of truth.
There are advantages and disadvantages to both approaches and organizations should choose the strategy that works the best for them.
Whether the enterprise decides on a single source of truth or a single version of truth, the focus on data quality has to be absolute. If decisions are going to be made on the data provided then the data must be trustworthy. The profiling of data should be done regularly to ensure that data has retained its quality. Alerts and thresholds on key business metrics should be set to ensure that nothing within the data is causing unwanted/unnecessary blips.
Data SLAs have to be clearly defined and communicated.
Data Governance and Data Privacy are key elements of the strategy. It is very important to understand who has access to PCI/PII/insider data and controls have to be put into place to ensure that access is managed to meet the guidelines set forth. Data Governance also requires that access protocols, archival and retention policies, SLAs, data lineages, enterprise KPIs and metrics in a business glossary, clearly defined and documented and easily accessible by all within the organization.
Monthly/Quarterly stakeholder meetings for communication of progress/status of efforts. This is very important to have continued support and success to see the initiative through.
Now that we have spoken quite a bit about the strategy itself, let's talk a little about what's required from the end-user’s standpoint to make this a success.
The largest obstacle is culture. You may have the best data strategy, execution, and environment available, but if you cannot get the right audience to use your data for their decision making, it is a lost cause. The top-down approach where the directive comes from executive management to use a platform works in some cases, but for people who are used to a legacy way of doing things, it becomes difficult to break them away from what they are used to. In this case, it is very useful to find data champions/power users within the organization and use the bottom-up approach to serve your purpose. Again, frequent meetings and communication with these data champions/power users have been proven to be a very effective strategy.
Tools for users to access your data easily and seamlessly is as important as building the platform itself. You may need to support an array of tools rather than enforcing just a single tool. For example, for reporting, you may need to provide a traditional BI tool, a data visualization tool, and a tool that supports even advanced analytics. It is imperative, however, that all these tools use the common data platform used so that the outcomes are consistent irrespective of the tools.
Training should be provided on the data platform and the tools periodically to ensure quick and continued adoption. Feedback from users should be understood and implemented effectively.
As with any initiative/project, key success metrics need to be defined and measured to ensure that it is providing the ROI it was meant to provide. A key success metric for analytics-driven platforms is time to insight which helps to measure how successful you were in making sure that everyone within the enterprise has access to the data and how quickly they were able to perform the analytics/ analysis they needed to.
The above considerations in your enterprise strategy should be able to guide your organization to be analytics-driven.