Functional Specification Documents are produced by the Business Analysts and document what the BI applications should deliver to bring value to the business. They should define the requirements for the reports, dashboards and analytical capabilities. For ad hoc analysis, the functional specification should further define the data items, their relations, hierarchies and how they will be used to perform ad hoc reporting and analysis.
Use Case Models and User Stories are good methods of articulating the business requirements. A user story written for the business users to understand and helps define the context of the functional requirement definition. Use cases are more detailed, less ambiguous and are directly consumable by a developer.
The Requirements Matrix catalogues all the required functionality defined in the functional specifications, assigning a priority to each item.
The Data Dictionary, also called the metadata catalogue, provides definitions of all data items that will be contained within the data warehouse. The aim of the dictionary is to be a single point of reference for all project members so that a clear and common understanding of the data is achieved across the team and to the business.
The training strategy documents how the business users will be trained on the new system. This strategy considers different user profiles and decides on the training method: classroom, online, peer-to-peer.
These are the documents, presentations, demo systems, demo data required to deliver the training.
User Guides describe how to access the system, where to find reports, dashboards and other BI content. For ad hoc reporting and analysis the user guide describes the semantic layer, the data objects and how the data objects can be used to perform reporting and analysis. Good user guides are heavy on examples and are kept up to date with changes in the data warehouse once it has gone live.
Source System Data Models or Entity Relationship Diagrams
The source system data models document the data structure of the source systems, the relationships, data hierarchies and data life cycle. Ideally these exist prior to the BI project but often don’t. If these are being produced for the BI project they are scoped to only include the data items relevant to the BI project rather than the whole system.
Interface Specifications document how the data is to be transferred from the source system and provided to the data warehouse. The data is either pushed to the data warehouse or it is pulled from the source system. In either case the interface specification is a contract between the two systems that define what data is being transferred, data structures, transfer method, frequency, schedule, estimate of volumes and security considerations.
Interface Specifications are typically produced by source system experts but can also be produced by the data integration developer.