Business Understanding
Identify each stakeholder and their objectives
In this initial phase, the focus is on clearly defining the objectives of the business intelligence project. This involves identifying stakeholders and business activities to determine what falls under management control or strategic oversight.
Management control focuses on monitoring and managing daily operational performance within an organization. It is an internal process aimed at ensuring that the company’s activities align with predefined plans, objectives, and standards.
Strategic oversight, on the other hand, concentrates on defining the company’s major directions and long-term objectives. It is a comprehensive approach guiding the company in achieving its mission and long-term vision.
By clearly identifying the purposes of the Business Intelligence project, you will be better equipped to understand user requirements.
Example: If you work for an online commerce company, the goal might be to measure online order processing times and identify bottlenecks (management control) or compare the performance of online sales with physical store sales (strategic oversight).
Specify requirements
In this step, you will specify the requirements of end users for your dashboards. The initial step is to identify relevant business processes for our project. For each of these processes, we need to determine the dashboard’s objective and the reasons for its creation. In other words, we need to answer the question, “What is the purpose of this dashboard and how will it be used?”. One result of this stage is the production of user stories.
Definition of the solution
Define the indicators
In this stage, you need to define the indicators that will be used to measure the performance and efficiency of the business processes. There are different types of indicators: performance indicators, efficiency indicators, insight indicators and trend indicators. Each of these indicators has a specific role to play in analysing performance and business processes. Some indicators are simple to define on the basis of business needs and data, and will enable a first iteration of the dimensional diagram to be produced. Other, more complex indicators will require the creation of more complex aggregates or calculations combining several other indicators.
Design dimensional schema
Dimensional schema design focuses on decision-making data by relying on business processes. It generally adopts an approach where the schema is structured in a star, organizing information into fact tables surrounded by dimensions. These dimensions describe business entities such as products, customers, suppliers, and dates, while facts contain measures related to business events. The granularity level of the fact table and conforming dimensions are fundamental concepts contributing to the consistency of historized data.
Example: If your goal is to analyze the revenue, costs, and profit margins of your online commerce company, dimensions such as date, product categories, and sales channels can be defined. These dimensions reflect the specific business context, allowing data to be organized meaningfully. For instance, the “product categories” dimension may represent the classification of products in the inventory management system, while the “sales channel” dimension differentiates revenue sources, whether online, in-store, or through partners. Measures such as revenue, costs, and profit margins are stored in the fact table, which is linked to dimensions by foreign keys.
Design dashboard
Dashboard design requires thoughtful consideration, integrating two crucial aspects. Firstly, it is essential to think about the overall architecture of dashboards, considering how they will articulate to fit the decision-maker’s thought process. This includes how different dashboards will be interconnected, allowing a smooth transition from one analysis to another, providing a comprehensive and coherent perspective. Simultaneously, the structure of a dashboard itself must be carefully thought out. It is imperative to create a layout that promotes seamless analysis, eliminating unnecessary distractions to enable the decision-maker to focus on clear and effective information.
Example: In the context of our online commerce company, a first global dashboard could offer an overview of performance. It would include indicators such as total revenue, monthly sales trends, and the distribution of sales by product category. This design would provide a strategic view, helping adjust marketing strategies and identify top-performing products.
On the other hand, a second operational dashboard could focus on more specific elements. For example, it could present the conversion rate by sales channel, current stock levels, and order processing time. This dashboard would be designed to address more operational concerns, allowing for effective day-to-day resource management, stock optimization, and quick issue resolution.
Definition of processes and tools to produce the solution
Define ETL processes
The ETL (Extraction, Transformation, Load) processes constitute a crucial step in the Business Intelligence cycle, responsible for integrating data into the decision support system. In this phase, it is imperative to precisely define these processes by describing how data will be extracted from different sources, transformed to ensure its quality, and aggregated as needed.
Define the tests
At the same time, developing a test strategy is essential to ensure the reliability of data throughout the process.
Defining and planning tests is a crucial step in the Business Intelligence cycle and is necessary to ensure that data is correctly extracted, transformed and loaded into the BI information system. The aim of the tests is to detect and correct any faults or errors that may occur from the loading of the source data to the display of the indicators within the dashboard.