Say goodbye to proliferation of departmental BI

Companies have actively discovered many new ways to use their data assets for decision support, operational reporting and process optimisation. However, there is a fundamental shortcoming in most BI tools today in that they cannot support the full range of BI functionality in a single architecture - leading to excessive costs, excessive delays and excessive user dissatisfaction.

What is needed now is a BI platform that can deliver all functionality requirements in a single architecture.

Since the early 1990s, BI applications have evolved dramatically and in many directions as companies` access to, and appetite for, information has grown exponentially. Companies are demanding more ways to report on and analyse data. The dramatic expansion of data warehousing combined with the widespread adoption of enterprise applications, such as enterprise resource planning and customer relationship management, and the overall increase in computer literacy, have fuelled this exponential demand for BI reporting and analysis applications.

In technology-savvy companies, departmental BI tools are being replaced by one industrial-strength BI platform that can deliver all five styles of BI.

Charl Barnard, GM, MicroStrategy

BI technology vendors reacted to the BI evolution the way software vendors always react to new evolving markets - by building niche software to implement each new pattern of application that companies invented. These patterns of applications resulted in software products centred exclusively on one style of BI - each representing a certain characteristic end-user function.

1. Enterprise reporting: Broadly deployed report formats for operational reporting and scorecards/dashboards targeted at information consumers and executives. These are destined for broad distribution to many people.

2. Cube analysis: OLAP slice-and-dice analysis of limited data sets, targeted at managers and others who need a safe and simple environment for basic data exploration within a limited range of data.

3. Ad hoc query and analysis: Full investigative query into all data, as well as automated slice-and-dice OLAP analysis of the entire database - down to the transaction level of detail if necessary. This is targeted at information explorers and power users.

4. Statistical analysis and data mining: Full mathematical, financial and statistical treatment of data for purposes of correlation analysis, trend analysis, financial analysis and projections. Targeted at professional information analysts.

5. Alerting and report delivery: Proactive report delivery and alerting to very large populations based on schedules or event triggers in the database. Targeted at very large user populations of information consumers, both internal and external to the company.

The end result of departmental BI tool limitations and departmental BI purchasing has been the proliferation of islands of BI that we see today. However, departmental aspirations must give way to company requirements. In technology-savvy companies, departmental BI tools are being replaced by one industrial-strength BI platform that can deliver all five styles of BI.

There are five macro forces obsolescing the "strategy" of isolated departmental islands of BI and the use of disparate departmental BI tools:

1. Enterprise-level BI applications need to access more data and support more users, while departmental BI lacks user and data scalability. Most companies have captured the low-hanging ROI fruit with their current array of departmental BI applications. They are now emboldened to move to the next level. And that means delivering much richer reporting and analysis, from much larger pools of data and delivered to many more users. Unfortunately, departmental BI tools today cannot scale to these new levels. The very nature of their underlying architecture prohibits them from analysing terabytes of data and delivering to expanding user bases. Companies need an industrial-strength BI platform designed for scale to replace departmental BI tools designed solely for interface functionality.

2. Inconsistent versions of the truth are propagated across the company - multiple islands of BI result in many, inconsistent metadata repositories. When the number of BI applications achieves a certain critical mass in the company, there is an inevitable overlap in analytical and reporting domains. Multiple reports from multiple independent BI applications present inconsistent information. As the number of applications increases, these inconsistencies undermine the integrity of all the BI applications. Companies need a BI technology that is built on one central, unified metadata that ensures consistency across hundreds of BI applications and the thousands of metrics, attributes and filters that make them up.

3. Users are increasingly dissatisfied about being forced to use multiple BI tools. When there are few BI applications, any given person only uses one of those applications. As the number of BI applications increases, more and more people will be accessing multiple BI applications, and hence using multiple different BI user interfaces to view reports and manipulate those reports. This means that BI users need to learn different ways to do everything, including such common actions as finding reports, running reports, scheduling reports, editing reports, saving reports, sharing reports, answering prompts, sorting the data and pivoting the data. With two tools, users will face challenges. With four or five tools, this becomes a valid reason for user rebellion. Nonetheless, this is the situation in which most companies now find themselves.

4. Enterprise-level BI applications require a richer user experience encompassing multiple styles of business intelligence - single-purpose BI tools cannot mix and match styles of BI. However, there is really a natural user flow starting with information presentation, to problem isolation, to full investigation, to advanced analysis, to proactive tracking and alerting. Technical boundaries that isolate the styles of BI are purely artefacts of single-purpose technology, and have nothing to do with user preferences or natural application boundaries.

5. IT organisations cannot afford the excessive cost of managing multiple BI technologies - disparate BI technologies for multiple BI applications are burdensome. With multiple BI tools, companies need to train people in the development and operational intricacies of each BI technology, establish specialised technical support teams, manage contracts with each BI vendor, and go to conferences, user groups and support forums for each BI vendor. Only with a single BI architecture can a company avoid all these redundant costs and efforts.

The ideal BI architecture is one that can deliver:

1. Any or all styles of BI which can be mixed and matched seamlessly for end-users, where the addition of each new BI style adds functionality to users` existing reports.

2. All expressed through a single unified user interface to maximise ease of use and user acceptance.

3. All delivered on top of a single integrated backbone that unifies the metadata, security and user profiles, ensuring a single version of the truth throughout the enterprise and thus minimising administration and maintenance efforts by IT.