Self-Service BI Tools

Self-service BI is a term for business intelligence (BI) tools aimed at business users, instead of the IT department. These tools help analysts explore databases through visual interfaces, instead of SQL queries and custom scripts.

We'll take a look at the types of self-service BI tools on the market and explain how they differ from traditional platforms. Here's what we'll cover:

What Is Self-Service BI?
Common Functionality of Self-Service BI Tools
What Type of Buyer Are You?
Benefits and Potential Issues
Market Trends to Understand

What Is Self-Service BI?

To answer this question, you have to understand traditional BI.

Back in the 80s and 90s, BI platforms were so complex that IT staff had to help business analysts create custom reports by writing SQL queries and custom scripts. Moreover, organization-wide reports were typically produced by the IT department.

IT was traditionally in charge of BI because traditional BI platforms store data in a data warehouse, which is a dedicated database system for historical business data (e.g., many years' worth of sales data, accounting data, customer information etc.).

Essentially, data warehousing involves pulling data from business applications (CRM systems, accounting systems etc.) and storing it so that analysts can spot and diagnose issues with the business's operational and financial performance:

Before data could be loaded into the warehouse, however, IT departments had to prepare it for analysis by normalizing dimensions (e.g., ensuring that “customer ID" means the same thing in all the tables in the database), aggregating certain metrics, cleaning dirty data etc. This is known as the “extract, transform, load" or ETL process, because data is extracted from applications, transformed into standard formats for analysis, and loaded into the warehouse.

When an analyst had to combine a new source of information, such as a web analytics system, with data in the data warehouse, the IT department would often have to reinvent the whole ETL process.

This was definitely not a user-friendly situation for analysts, and the IT department didn't really like being reduced to the work of data waitressing either.

Common Functionality of Self-Service BI Tools

Self-service BI tools emerged in the last decade as a response to this situation. They differ from traditional systems in the following ways:

• Direct/live data connections. With a traditional BI platform, data is loaded into a warehouse before analysis. With a self-service tool, analysts simply connect directly to the data source, whether it's a cloud application, a relational database, a NoSQL file system, a flat file etc. Analysts can then explore and combine data from the data source with other sources. Moreover, instead of pulling data from the source in batches, analysts can configure a "live" connection to the source so that dashboards and visualizations automatically refresh with new data.
 
• Self-service data preparation. Instead of having the IT team clean and prepare data for analysis, self-service tools allow analysts to prepare data themselves. After the analyst connects to a raw data source, they can specify cleaning procedures and transformations that the data needs to go through before being presented in, say, a dashboard.
 
• Self-service data modeling. As analysts clean and transform data from a raw data source, a data model emerges. For instance, the analyst decides which columns in which tables should be aggregated using the “customer ID" dimension. Essentially, with traditional BI systems IT settles on a data model before data is loaded into the warehouse, whereas with self-service BI business analysts create the data model as they clean and transform data sources for analysis. Pre-configured data sources can then be shared with other analysts for use across the organization.

Self-service data modeling in Looker BI

• Graphical user interface. Self-service BI tools include a graphical user interface for data analysis. Analysts can drag and drop dimensions and measures into charts to create data visualizations. They can then click on parts of the visualization (e.g., a wedge in a pie chart) to drill down to more granular data. The drag-and-drop gestures that analysts make in the user interface are transformed into SQL queries by a backend engine. This means that analysts don't need to know programming languages to get data out of databases.

Visual data exploration in Board

What Type of Buyer Are You?

Analyst workgroup. Analyst workgroups typically need solutions that are robust when it comes to data analysis features—for instance, the statistical algorithms found in IBM SPSS and SAS STAT. They don't typically need strong dashboard capabilities or governance functionality.

IT department. Even though self-service tools are designed for business users, IT departments are still tasked with purchasing and configuring these platforms in many organizations. IT departments will need to find solutions with strong data modeling capabilities, since they'll need to invest in tools that can support data modeling for a whole organization rather than just a workgroup concerned with specific analytical tasks. Moreover, they'll need to invest in solutions that offer strong data governance to protect sensitive data from business users who don't need to access it as part of their roles.

Benefits and Potential Issues

Self-service BI tools offer the same benefits as traditional BI platforms, including:

Data governance	Restrict access to sensitive data sources and visualizations.
Role-based dashboards	Customize end-user dashboards for different roles throughout the organization.
Data mashups	Blend data from many different sources in visualizations.
Metadata management	Manage metadata (time stamps, classification tags etc.) across the organization.

Self-service tools also offer the benefit of user-friendly, visual data exploration, which isn't a strength of either traditional BI platforms or of tools for statistical analysis and data mining.

The need that self-service tools don't cover is data storage for analysis, since these tools typically lack support for data warehousing.

Additionally, since self-service tools lack support for analytical data warehousing, they are not as robust when it comes to data governance as traditional platforms. Organizations in which data quality is of utmost importance (e.g. financial institutions) will still need to use traditional data warehousing alongside self-service tools.

Market Trends to Understand

• The convergence of data prep and visual analytics. Self-service BI tools fall into two categories: data preparation tools that allow analysts to clean and prepare data for analysis, and visual analytics tools that give analysts a visual interface for data exploration and enable them to publish visualizations as dashboards. Currently, data prep vendors and visual analytics vendors team up for self-service BI deployments. However, visual analytics vendors like Tableau are now working to add data prep functionality to their products. This means that you should consider your vendor's roadmap for product development when selecting a self-service tool.
 
• Analytics as a service. As we've seen, self-service tools lack data warehousing functionality, which remains necessary for many organizations. However, providers of cloud services like Amazon Web Services (AWS) are stepping in to meet this need. Services like Amazon Redshift and Amazon S3 are enabling businesses to shift data warehouses to the cloud, and Amazon is launching a whole line of cloud-based analytics tools to complement data storage offerings.
 
• “Smart" data discovery. New analytics offerings like IBM Watson and Salesforce Einstein promise to revolutionize analytics by automating the data exploration process via machine learning and natural language generation (NLG). These tools support a Q&A interface for data analysis, and use machine learning algorithms to spot patterns in data on their own, rather than relying on human users to spot the patterns. NLG functionality allows smart data discovery tools to return answers in the forms of sentences and even short narratives (“EMEA Sales dropped in Q2 because of APAC supply chain interruptions"). While smart data discovery is a long way from replacing human analysts, it represents the future of self-service analytics.