On a recent business trip to St. Louis, MO, I happened to read Zen and the Art of Motorcycle Maintenance. One quote from the book that particularly appealed to me was: “We have artists with no scientific knowledge and scientists with no artistic knowledge ... and the result is not just bad, it is ghastly.” I have seen examples of this premise and think the reason I can relate to it is simple: managing projects in an iterative workstream-based environment is equal parts art as it is science. Let me explain.
First, the science. The deployment project for a next-generation database solution at Merkle is comprised of six workstreams:
- Data Management — this workstream involves modeling a client’s data entities and relationships onto logical and physical data models (think blueprints for your house’s construction project).
- Data Integration — the 800-pound gorilla (or tuna if that floats your boat) in the room and accounts for almost 3/4th of the development effort in any database project (think wiring, plumbing, masonry, and everything in between). While this workstream can be kicked off as soon as source system extracts (data inputs) are finalized, the data integration and data modeling workstreams need to be tightly joined at the hip for a quality end product.
- Campaign Management — covers the setup and configuration of a campaign application (IBM Campaign, RedPoint, and SAS are the usual suspects) and initial setup of a handful of marketing campaigns.
- Business Intelligence — involves the setup and configuration of a BI tool like Tableau, MicroStrategy, or Cognos. Marketing performance dashboards and reports are also typically prototyped during the project and transitioned to the client.
- Connected Recognition — Merkle’s secret sauce for Customer and Digital Data Integration, this workstream typically runs parallel to the above four during the design and development phases of a project.
- Infrastructure — this workstream involves the procurement, installation, and configuration of all the hardware components needed for the next-generation database solution.
While the durations of these workstreams depend on any given client’s business and IT requirements, two things never change in every engagement — the fact that these workstreams are inter-related and that they mature iteratively through the course of a project. Here is my list of top three inter-relationships that need to be managed:
- The data modeling workstream is dependent on upstream discovery activity where client’s campaign and BI requirements are captured and data inputs are documented.
- The data integration workstream is contingent on data modeling of the warehouse and marketing database being complete. Typically, one should expect the data model to go through not more than three iterations, and it is natural to have some downstream impact to the data integration workstream on account of these changes.
- The BI workstream is dependent on the data model being instantiated (tech speak for saying that the target tables are created in a database and seeded with sample data) in time for iterative prototyping of reports and dashboards.
Where is the art in all of this, you ask? The creativity really is around the method that the team employs to iterate between the workstreams. For example, if more than three iterations are required to finalize the data model, then the team runs the risk of creating a ton of downstream rework on the data integration workstream, resulting in schedule and budget overages. The secret sauce here really is the team’s appreciation of the client’s business and a thorough understanding of the client’s customer data.
At Merkle, we call the art and science of managing workstream-based projects the MerkleONE methodology. Visit the MerkleONE Delivery Methodology page to learn more.