You use business intelligence to monitor and investigate various aspects of your company – sales, marketing, operations, and others. But as a data-driven organization, you should put your business intelligence tools and initiative under the same scrutiny. This means using the same techniques and tools with which you measure every other aspect of the business to measure the effectiveness of your BI itself.

This article will present some of the core goals and concepts of ‘BI on BI’, as well as the metrics you should track to evaluate your current BI implementation according to various criteria.

Typical Goals and Dashboard Consumers

This type of ‘introspective’ analytics will generally be focused more on operational, rather than analytical reporting. The typical consumers of these reports would be the project owners (CIO \ business analyst \ business executive \ other, depending on the organization). Generally these dashboards aim to:

  • Monitor the health of the BI implementation
  • Understand how business intelligence is being used in the company and which KPIs are being tracked
  • Identify redundancies and performance issues
  • Gain some concept of the current ROI you are receiving from your business intelligence
  • Find ways to improve the above

In effect, you’re monitoring two different aspects of the BI deployment: system usage, i.e. which dashboards and queries are frequently being used; and data usage, i.e. which sources, tables, and columns, are being processed on a regular basis.

Note that all of this is more relevant to larger-scale deployments – if you have only a handful of users and 2-3 dashboard reports, BI on BI would probably be overkill.

Wait… where’s this data from?

Obviously, if your current business intelligence is being managed via a series of disconnected Google or Excel spreadsheets, CRM dashboards and reports and other disparate BI systems, this type of analysis would be extremely difficult if not impossible. If you’re already using proprietary BI tools, but your business intelligence is spread between separate ETL, database and visualization platforms, aggregating all the data could also prove difficult.

Since Sisense is a Single-Stack™ BI tool, you can easily access all the relevant information regarding user actions, data being processed, performance, and dashboard usage, which is extracted the following tables:

User data from Sisense Elasticube

Let’s proceed to see how we can use BI on BI to understand our current implementation’s strengths and weaknesses:

Monitoring Adoption Rates

When you decided to implement a new data analytics platform, your initial BI project plan probably included the departments or people who would access to the system, and how this would fit in with their overall goals and objectives.

To ensure that the actual implementation is in line with these plans, you should regularly track:

  • Which individuals, departments or branches are actually logging in to the BI platform?
  • Which dashboards they are viewing and which data is being queried?
Actions by dashboard bar chart

Thus for example, while it’s perfectly natural for the VP Sales to only log in a few times a day to see her team’s current status, if the resident data analyst – who should be your power user – is not using the system, this could indicate a problem with your implementation or training.

Data Sources Being Processed

You can use BI on BI to see:

  • Which data sources or are being accessed more frequently?
  • Which specific tables or columns are being used?
  • Who is using these data sources, tables or columns?

This can help you identify cases where you are importing redundant data, which might unnecessarily slow down performance. Additionally, if some of the data is being extracted from pay-per-use cloud sources (such as AWS), you might identify an opportunity to save costs by limiting the data you import according to the actual usage in your company.

Types of Analytical Usage

Even if adoption is at the level you would expect, it can still be useful to understand how the BI platform is actually being used on a day-to-day basis. Understanding which types of analyses or reports are more popular in your company will help you consider whether the usage is in line with the ways your original intentions for the business intelligence projects, e.g. – are dashboards being used for deep data analysis, or are they being (under-)utilized strictly for operational reporting?

Analytical actions trend

Licensing Evaluation

An important factor you should be looking at is whether the people who have access to the system are actually using it, and at what frequency. This can help identify whether your current licensing structure is in fact the ideal fit for your needs, or if you can remove or change licenses (for example, replacing administrator permissions with viewer permissions). This also adds an additional layer of data security by enabling you to identify users who are viewing dashboards or data which they should not have access to.



Finally, system administrators or IT users can use BI to investigate redundancies that could potentially hinder system performance. Keep daily tabs on dashboard and widget response times, and locate specific components which might be slowing you down – either due to poor data modeling, excessive data being queried, or other solvable issues.

But don’t stop there!

These are only a few of the stats you might want to track. Once you have your basic BI-on-BI dashboards and data model in place, use it to further investigate your implementation. Drill down into specific dashboards, users or data sources. Ask ad-hoc questions to find out more about what business intelligence is currently doing for your organization, and to recognize potential improvements to make or redundancies to eliminate.

Tags: |