An important thing to consider when thinking about how organizations can get started with advanced analytics is how to structure an analytics project. The four basic phases of typical data analytics projects are described below.
Phase 1: Problem definition or business need identification
Conduct a work session with the client/stakeholder and define the problem or business need. Develop an understanding of the related business processes and what types of data collection are necessary to answer the business problem. Can you rely on historical data or will you require new data sources? Define what constitutes a successful project, but avoid problem solving – focus on what is causing “pain” to the client. This is the part of the process where you will learn the information you need to tie your analytics strategy to a business case.
Phase 2: Agree on requirements, scope, assumptions, project value, champions, and partnerships. Draft a project charter.
Once the project has been identified as a viable project for your organization, focus on defining requirements, scope, and assumptions. This exercise should be done with the client/stakeholder. Also work on defining project value and prioritize the project based on value, feasibility, and impact. Define project champions and partners. An output of this step could include the drafting of a project charter. This stage is meant to foster agreement between data scientists and business users about what new analytics capabilities can be developed and what business intelligence purposes they will serve. Once the analytical teams and business teams are speaking the same language, we can move onto Phase 3.
Phase 3: Prototyping and development – Define modeling approach, technology design architecture, and metrics of success
Assemble a development team (including analytics professionals, software engineers, and other staff as required). The team constructs, tests, and refines a system prototype while frequently interacting with prospective users. Recommend changes in processes and procedures necessary for effective system performance. Once the prototype is successful, move to full development and deployment. Successful analytics leaders constantly provide updates about a project's progress and any unexpected hurdles during this period.
Phase 4: Implementation and maintenance plan
Install the system, train operators, revise processes and procedures, provide for maintenance and future upgrades, and measure benefits. Analytics tools frequently need tuning over time, so make sure you don't neglect this step.