How to implement a Business Intelligence (BI) system in your organization
Implementation of Business Intelligence solutions increases the organization’s capacity to understand its data and make right, data-based decisions. ‘5 reasons to implement Business Intelligence for data processing’
Assuming you have already taken the decision to implement a BI system in your organization, this article will help you to take the first step into the implementation.
Business requirements
Start by understanding your environment and the customer experience before choosing a technology.
In my case, these are the requirements:
- Provide easy and secure self-service access to reports
- Host several data sources in a secure manner
- Allow advanced users and BI admin to create and maintain existing reports
- Allow BI admin to create and maintain existing data sources
- Ensure seamless business continuity, without requiring extra maintenance
- Ensure seamless integration with internal systems (SAP CRM, MS-SQL, etc.)
Benchmark BI systems
There are many BI solutions in the market, check the Gartner quadrant to get an idea.
In my case, I was between Tableau and Power BI and the selection was difficult… finally I chose Power BI mostly because it’s compatible with different Microsoft services, such as office 360, Azure Active Directory, Power Automate, OneDrive, and other Azure services (cloud storage, AI and ML, etc).
Again make sure the BI system selected cover your business requirements and the budget.
Implementing your new BI System
Users (roles & permissions)
Users will have different roles and permissions depending on their needs and skills. A general approach:
Viewers: end-users with ‘read-only’ access to reports.
Members / Contributors: advanced users with ‘read and write’ access to reports and datasets.
Admin: can add and remove users, publish apps, and all other options.
Workspaces
The PBI service is organized in Workspaces that host Reports, Dashboards, Datasets, and Dataflows.
In Power BI, there are two types of workspaces that offer different options: Pro or Premium.
Pro: for web reports with all basic functionalities you might need to create reports, datasets, and dataflows.
Premium: same as pro, but additionally offers Artificial Intelligence (AI) and Machine Learning (ML) functionalities provided by Azure inside PBI. Also, the premium capacity allows the creation of Paginated reports, here an article to learn more about paginated reports.
The trick here is that users accessing these workspaces will also need the same type of license.
Data Sources
List of all the data sources that will be queried, these can be databases, logs files, websites, etc.
Data stewardship plays an important role at this point, it should be defined data ownership, security, access (public/private), etc.
Create a centralize repository
In cases where there are complex systems with many data sources, it would be better to implement a data warehouse to facilitate data cleaning and accessing, in this way you can easily write queries that will answer your business questions.
In Power BI, there are Datasets and Dataflows available, this could somehow play the role of a data warehouse (for simpler systems).
Dataset: each report has a dataset associated (normally with the same name), this is maintained by the report owner.
Dataflows: similar than a dataset but users can create their own dataset version from it (without modify the dataflow). This is recommended when many reports need to access and modify the same data model, so this model works as the ‘main source of trust’. Dataflows are maintained by the admin.
Automatic refreshes of data sources
Once you know where your data is located, you might need to setup a synchronization system to update the data in the reports. In Power BI this refreshes are done using a gateway.
Internal data
Both datasets and dataflows have the option of ‘automatic refreshes’ through a PBI gateway, access to data sources is done in a secure manner, credentials are managed by the gateway.
Refreshes can be scheduled as frequent as once an hour to once a day, depending on users’ needs. In case of failure, an email notification is sent to the data model owner.
Cloud data
For data stored in Microsoft services, such as: OneDrive, MS 365 groups (Teams / Sharepoint), Azure cloud services, etc. data is automatically updated in real-time using MS secure authentication mechanisms, no gateway required.
There will be more or less considerations depending on the complexity and size of the systems, but in general this should give you a good start to implementing your BI Systems.
