A Data Strategy Playbook

My playbook emphasizes speed and consistency through simplicity, pragmatism, and prioritizing decisions over deliberations.

Here's a summary.  I'll dig into the details in future posts.

Phase 1 - Draw a Map.  I collaborate with the client's decision makers and check writers to articulate a clear destination for their data aspirations.  Then I talk to everyone else to determine how far they are from achieving them.  Finally, I define reasonable milestones that map a realistic journey.  No milestone should take more than a month to reach from the previous.  Keeping the point-of-next in sight is key to avoid client fatigue and disillusionment.

Phase 2 - Develop rapid ingestion capabilities.  Nothing is more critical than this to a data program's success.  Data products require raw materials and a reliable supply chain to maintain ready inventory.  Any newly identified data source should take no longer than a week to extract, land, and historize on a defined schedule.  This step should include basic metadata and lineage capabilities.

Phase 3 - Prioritize Master and Reference Data.  These data sources should be among the FIRST to be ingested.  I don't waste time on non-productive deliberations here.  ALL clients have Customers, Employees/Workers, Products, Vendors, Calendars, and a Chart of Accounts.  Most also have Locations, Business Units, and multiple Categorization Code tables.  I'll work with your key subject matter experts to identify high utilization subjects and the amount of overlap across existing systems, then create a raw inventory of all views to promote rapid reconciliation and efficient decision-making.

Phase 4 - Implement a Mesh or Fabric.  This is the second most important characteristic of a successful implementation of my playbook.  What good is rapid ingestion of raw, historized, source data if it's not immediately available to people who need it most and understand it best.

What is a mesh or fabric?  Simply put, it's a means of organizing trustworthy copies of your important data to optimize function-based access and self-service, while providing the foundation for product development and effective governance.

You'll find a lot of online debate about the mesh/fabric concepts of data architecture, but I won't waste too much time splitting hairs. The essentials to me are making data available at the "point of business" while ensuring enterprise-wide consistency in tactics and tools.

Phase 5 - Collaborative Curation.  Once a viable inventory of raw data is created and "owned" by the right people, I facilitate conversations to optimize cross-functional use and sharing.  Sales needs Product details.  Everyone needs Employee details.  Decision rights and authorities are defined by proximity, and disputes escalate quickly until a decision is made.  The goal is to limit endless debates over functional data ownership and drive quickly to decisions. Data's value has a shelf life, and no one benefits from cumbersome exercises in governance.

Phase 6 - Rinse, Repeat, and Innovate.   Leverage the inventory of raw data assets mercilessly.  Within the first quarter, the client will have a foundation of essential data capabilities, a consumer base highly engaged through increased access with limited latency.  Increased freedom to explore, experiment, refine semantics, context and boundaries creates a value chain at all levels of the organization.  This is where AI opportunities will abound, but I will always recommend a solid foundation of understanding before augmenting with AI.

Soon, those dashboards you craved yet seemed so far away?  They will become cheap, disposable commodities as your organization moves ever closer to conversational business intelligence.