Analytics teams are actively working to redefine their technology stacks to meet the demands of modern big data and stay competitive. With technologies and best practices evolving at an incredible pace, there is no single blueprint for what is “world-class.” That said, my customers frequently ask me,“What are your other customers doing?”
After years of helping teams build modern analytics programs, we know not only what our customers are doing, but what they are doing well and how they are doing it. We decided to compile this information and develop a best practices data and analytics maturity model. This started off as an internal AtScale publication describing the value of a semantic layer to help create better analytics solution designs with our customers. However, we’ve received such positive feedback that we are now making the framework available to the public to help teams assess their current capabilities and plan where they should focus their future investments.
The Data And Analytics Maturity Model
The model framework is based on six capabilities that are core to building a data and analytics practice. The model looks at four levels (0-3) corresponding to your level of maturity. Organizations rarely are at the same level of maturity for all capabilities.
The first capability is related to data infrastructure and how it is managed. In less mature organizations, data can be siloed across multiple locations and formats (i.e., data marts, SaaS applications, operational databases and raw files). More mature organizations have moved toward centralized cloud data platforms and are leveraging data sources beyond their own applications (i.e., third-party data sources accessed from Amazon Data Exchange or Snowflake Data Marketplace).
Key Takeaway: When it comes to data, think outside the (corporate) box.
Next is access. Less mature organizations often leave users to wrangle their own data with one-off imports or extracts. More mature organizations create data pipelines using ETL/ELT data integration approaches to move data to a central location, pre-positioning it for users to access. The most mature organizations may incorporate data virtualization to provide live access to data without complex data movement and give users the choice of data access best suited to their particular use case. By applying DevOps methodologies of CI/CD (continuous integration/continuous delivery) to their data operations, “DataOps” provides the most mature organizations with the agility to respond to dynamic business requirements. Finally, the most mature organizations actively share data with suppliers and partners to drive insights across their value chain.
Key Takeaway: Live data access and DataOps deliver analytics agility and speed.
Data modeling is about making data consumable by ordinary humans. Less mature organizations ignore data models and default to building one-off datasets that too often produce inconsistent results. Most organizations rely on tabular data modeling with data managed within data warehouse tables. The tabular approach provides some standardization but still leaves the interpretation of raw data up to the data consumers. The most mature organizations define a logical, dimensional model of their data to expose a consistent, business-friendly interface of key metrics to data consumers, making data accessible to a wider range of users.
Key Takeaway: Dimensional data models make data accessible to everyone, not just power users.
Data consumption refers to how users query data and how they share insights with others. For less mature organizations, custom SQL and code are the primary tools for asking questions of data, limiting access to a small set of power users. More mature organizations provide “data as a service” to allow users to explore and report on business metrics using tools of their choice. The most mature organizations deploy “data as code” by embedding analytics inside applications and business operations to move beyond the dashboard or BI tool.
Key Takeaway: There’s more to analytics than SQL and canned dashboards.
Insights relate to how data and analytic teams empower better decision-making and the creation of new business value through the delivery of data-driven insights. Less mature organizations focus only on historical data. More mature organizations produce insights that predict future results (i.e., predicted sales and inventory) and prescribe actions to capture opportunities.
Key Takeaway: Look to the future, not just the past.
The Data and analytics maturity model is a tool to chart a path toward improving overall organizational capability. Organizations typically do not fit neatly into one of the four maturity levels. It’s more common to span different levels of maturity according to your organization’s capabilities. In future posts, I’ll discuss strategies for advancing your organization’s maturity along each of the six data and analytics capabilities.