5 Things Most Business Intelligence Applications Need to Do Better

5 Things Most Business Intelligence Applications Need to Do Better

In the current Business Intelligence (BI) space, there are several key trends and features that seemingly everyone is looking to embrace.  Cloud and Big Data integration, mobile delivery, self-service enablement and advanced visualization will show up as requirements even when they are not all that relevant to the application in question.

On the other hand, there are five other key attributes that are relevant to nearly all BI applications that are not universally supported by the commonly used toolsets. Even when they are, they are often neglected by development teams. Here they are:

Real Security

There have been a number of high-profile data breaches as of late, including a massive theft from the health insurer, Anthem. That one was notable because it was not the result of a breach of a customer-facing application. It was a theft from a vulnerable, unencrypted data warehouse. Those of us who practice BI and data warehousing have always been evaluated on our ability to make more data available to more users with the least amount of effort. In the process, we tend to create ‘one stop shopping’ to those who would access proprietary data with malicious intent.

Data warehouses generally sit behind strong perimeter defenses, but the majority of successful attacks come from inside the firewall. This leaves them completely vulnerable to anyone with network access, unless the data is encrypted and/or there is strong additional security at the application level. This risk is generally underestimated by delivery teams and not marketed heavily by the tool vendors (although that is starting to change).  Adding additional layers of security can make data access less convenient and slightly compromise performance, but the economic cost of a mass theft usually more than justifies this when considered objectively.

User Experience

For many years, I was fortunate enough to work at a very popular consumer website and we had several expert user experience professionals on staff and available to us in IT. Early on, our BI applications were not being adopted as readily as we had hoped so I asked one of them to do a UX review of them. The results were not pretty. Many deficiencies were identified with both navigation and visual design. Some of these were a function of the tools we were using, but others could be addressed through configuration and minor customization. Once this was done, our adoption and user satisfaction improved considerably. We learned several lessons from this including:

  1. If you are deploying your BI applications across devices and screen sizes, do not assume the user cases will always be the same. Every situation has unique characteristics, but in general, tablets and phones are used to consume information, like scanning reports and dashboards. Larger fixed screens are used to customize displays, collaborate on analysis and create new analysis using self-service capability.
  2. Apply the same web analytics, A/B testing and survey tools on your BI sites that commercial sites use. They will often yield valuable insights on how to improve utility and user satisfaction with relatively minor adjustments to your code and configurations.


BI is all about decision support and decision making is generally a collaborative process in business. Curiously, BI tools and applications have historically offered very little in the way of support for collaborative analysis. This, if considered at all, has been relegated to embedding BI within portal tools like SharePoint or simply left to MS Office integration and the endless email threads that it can produce.


Data in context becomes information and metadata, or data about the data that we consume, often provides that context.  Data lineage, quality ratings, recency and business rules are all useful to know for proper interpretation of that data by consumers.

For developers, the technical metadata is necessary in order to integrate and report it properly. Tracking data dependencies is also very important. That is, all the places where that data is used or stored. Without this, it can be very difficult to identify and perform the necessary work in response to any upstream changes to the sources or basic characteristics of data being imported into the BI platform.

Usage Tracking

Although many of the more robust BI tools have provided the capability to track usage and govern resource consumption for some time, it is often not used by administrators for anything but shutting down runaway queries and chargebacks. This “BI about BI” becomes a missed opportunity to gain important insights about what is working, what is not and where performance can be enhanced.

If you need an example of BI and decision support applications that tend to get all of these things right, look no further than any of the consumer facing sites from the largest retail investment firms. They are meant to provide all the necessary information to support investment decisions. Out of necessity, they are secure and can update in real time. They are constantly measured, monitored and experimented upon for continuous improvement of the user experience. As a result, these sites are intuitive enough to use confidently without the need for training or user manuals although online help and metadata is there when you want it. They integrate structured with unstructured text for context, and provide great flexibility and user control over the presentation of tables and visualizations. They even support collaboration with their social features. Most importantly, they make the data actionable immediately.

Can your BI portal do all that? The tools and technology are available.  It’s up to us as practitioners to meet that standard for the decision makers we support.

Article written by Stephen Robinson
Want more? For Job Seekers | For Employers | For Contributors