What is business intelligence (BI)?
BI is a technology-driven approach to help business users make informed decisions. BI is a collection of tools, techniques, technologies, and methodologies to achieve the goal of informed decision making. A typical BI process would consist of gathering and consolidating data from a wide variety of data sources from within and outside the organisation, transforming and cleaning it, and then reshaping it into a usable format. Often, business users visualise the trends of various domain-specific measurements from this data and visualise these trends in the form of dashboards.
What is self-service BI?
Self-service BI is allowing the business-user to perform analytics on the data. Traditionally, the organisations used to have specialised BI teams who would gather user requirements for the dashboards, identify the required data, clean and transform it (if needed), and consolidate it to a central repository (often a data warehouse), and eventually build and commission a dashboard on top. By looking at the number of steps involved, you can imagine the time and effort needed to enable BI. The intuition behind self-service BI is to cut this process short and allow the end-business user to be able to quickly draw valuable insights from the data without going through a lengthy process.
What should an organisation opt- for, traditional or self-service BI?
There is no one size fits all answer to this question. It depends on the needs. Traditional BI indeed requires more resources, but if you need to put in place a certain level of governance, auditability, and traceability for compliance to specific standards or regulatory requirements or any other reason, then certainly a well-structured centralised BI team is the way to go for it.
On the other hand, self-service BI empowers the business user to draw their insights in a flexible and agile manner. However, this introduces the risk of losing trust in the data because everyone may come up with their version of the truth.
How could we avoid creating “alternative facts”?
You let them play with the data but play by rules! You need to create somehow up to a certain level the governed data sources on which users could do self-service BI. You can create a culture where you give the users flexibility and freedom of doing their analysis, but at the same time, you govern, and quality check the data on which such analysis is performed.
You do not need a centralised team of BI experts, a corporate data warehouse, and standard procedures in place. But at the same time, it is essential that you do not lose trust in your data. That is why, when you start doing it, put in due considerations to creating data marketplaces where you ensure that quality data is available. In a recent survey, master data and data quality are identified as the most important considerations while enabling BI in an organisation.
Which typical roles are involved in a BI team?
This is a hard one to answer! With the big-data buzz, the roles and the titles have completely changed. It is often not very clear to tell the difference between a data engineer and an ETL developer in many situations. But in a typical BI process, we would expect:
A business analyst to gather and document the business requirements An ETL developer to collect and consolidate the data into a central repository, and A data analyst to build the reports and dashboards.
Is self-service BI the new market trend ?
If we look at a traditional BI project, about 80% of resources go into what we call Extraction, Transformation, and Loading (ETL) data before it can be used to create reports and dashboards. The principal motive behind self-service BI is agility. Therefore, it is also vital to enable the business users, who often do not have technical skills to manipulate data, to have access to data and make it usable in a flexible and agile manner. To facilitate the exploration, cleaning, integration and then use of data by the non-technical users is called self-service data preparation. In the past few years, quite a few tools self-service data preparation tools have appeared, and I believe that such tools will be adopted more and more by the organisations in the coming years.
Does this mean that the traditional IT-driven ETL will become obsolete ?
Not really ! I believe that just like traditional and self-service BI, both IT-driven ETL and self-service data preparation will coexist. The traditional ETLs have their place as they exist in the first place to ensure stable performance, data quality, appropriate level of aggregation, integrity, and conformance to all governance policies. So when you need a precise, quality-controlled data, your ETL team can deliver that, on the other hand, if a business-user quickly needs to exploit data to identify business opportunities, they can use self-service preparation to gather the data and self-service-BI to analyse data.
Written by a BI/DWH practitioner in the insurance sector.