Register for "A Guide to Data Products: Everything You Need to Understand, Plan, and Implement" - Friday, May 31,1:00 p.m. ET

Part IV: Seven Keys to a United BI Environment

This is the four in four-part series on enterprise business intelligence (BI). 

Read Part I - The Battle for BI is Over: Now the Hard Part Begins

Read Part II - Ten Characteristics of a Modern Enterprise BI Tool

Read Part III - The Holy Grail of Enterprise BI - User Adoption

Organizations have been attempting to deliver enterprise business intelligence (BI) for more than two decades, yet for many, success is fleeting or entirely absent. This begs the question: why is BI so hard? Industry experts trot out a litany of explanations: business users don’t know what they want, making requirements impossible to gather; business analysts just want data dumps not BI tools or applications; BI tools are too slow, complex and inflexible; self-service BI is hard; data quality is difficult to manage; and so on.

Know Your Users. Yet, the biggest reason BI implementations fail is that BI teams do not know their business users. Not personally, of course, although that may be true. But categorically. Most BI teams have never conducted a complete inventory of business users and their information requirements. If they did, they would quickly realize that there is an enormous mismatch between BI users and the architecture supporting them.


The biggest reason BI implementations fail is that BI teams do not know their business users.


The reality is that there are two major types of BI users: casual users and power users. And despite what vendors say, these two groups have diametrically opposite requirements. Casual users want performance data delivered at a glance with options for drill down navigation, while power users want the ability to combine disparate data sets to conduct various types of analyses. In short, casual users want tailored, interactive reporting (or dashboarding), while power users want unfettered access to data to drive insights.

Given these antithetical requirements, it is difficult to meet the needs of both casual and power users in a single BI environment. The big mistake BI leaders make is thinking that one uniform environment is sufficient to meet the needs of business users. In reality, BI leaders must build dual environments—one for casual users and one for power users—to succeed with BI.


BI leaders must build dual environments—one for casual users and one for power users—to succeed with BI.


Two Worlds of BI

The first step to BI success is to recognize that there are two worlds of BI.  Top down BI addresses the needs of casual users who use information to do their jobs, while bottom-up BI addresses the needs of power users who are paid to analyze and crunch data on a daily basis. Each world of BI requires different tools, architecture, and data. (See figure 1.)

Figure 1. Two Worlds of BI

Image title

Divergent Architectures.  In top-down BI, the BI team delivers reports and dashboards to casual users that answer predefined questions based on a user’s role and the processes that they manage. The BI team collects and models the data, puts it in the data warehouse, and builds reports and dashboards that run against that data. 

In bottom-up BI, by contrast, power users explore data in the data warehouse and other systems to answer questions that arise in the business units and can’t be anticipated or answered with standard reports. The analysts create their own ad hoc reports using Excel or some other low-cost tool and publish these to their colleagues.

Divergent Processes. These two worlds also differ significantly in other ways. Building reports in the top-down world is time-consuming, costly, and complex, since the information technology (IT) oversees the process. But the data is consistently defined and the output is scalable, secure, and accurate. Business users use the same data and metrics to make decisions, fostering business alignment, which makes the CEO happy.

The bottom-up world is the opposite. Building ad hoc reports is quick, easy, and inexpensive, since a business analyst with Excel can do the job. This makes the line of business manager happy. But the output is not scalable or secure and the data is usually haphazardly defined and inconsistently defined.  These data silos undermine information consistency since everyone uses different data and metrics to make the same decisions.

Enterprise versus divisional BI.  In many ways, top-down and bottom-up BI mirror the tensions between enterprise and divisional BI deployments. Business people in the division want and need information quickly and don’t want to wait for the corporate BI team to deliver it. Therefore, they often embrace bottom-up BI. In contrast, the corporate BI team is hired to deliver scalable and secure applications with high quality, integrated data that is consistently defined. Therefore, they tend to embrace top-down BI and look suspiciously on bottom-up BI activities.  (See figure 2.)

Figure 2. Top down versus bottom up BI

Image title

Typically, the top-down and bottom-up worlds are at war with each other. The BI team locks down the data warehouse to prevent power users from issuing runaway queries that bog down performance. The power users, in turn, circumvent the BI team, and collect data on their own, creating spreadmarts or renegade reporting systems that undermine the value of the data warehouse. The BI team views power users as renegades, while power users view the BI team as a huge obstacle to getting work done.


Typically, the top-down and bottom-up worlds are at war with each other.


Reconciling the Two Worlds

Successful BI leaders reconcile the two worlds of BI and turn potential enemies into allies. Although analysis and reporting are flip sides of the same coin—analysis begets reports and reports beget analysis—uniting these two worlds is not easy. Below are seven tips for uniting the two worlds of BI.

1. Recognize the Differences. The first key, as mentioned above, is recognizing that two worlds of BI exist and must be treated differently. This doesn’t mean that BI teams can’t create a unified BI architecture. It’s just that they need to carve out space for each type of user within the BI domain and make sure they each has the right tools and data to support their business requirements.

2. Don’t Homogenize. Many BI leaders are tempted to homogenize the two worlds, not harmonize them. They force fit the organization with a one-size-fits-all approach to BI. For instance, giving casual users tools or data designed for power users always backfires: casual users stop using the tool and data and revert to submitting requests to IT for custom reports. Conversely, giving power users tools and data meant for casual users frustrates power users and makes them hesitate to work with IT. 

3. Sandboxes. To keep from driving power users underground, BI leaders need to carve out spaces in the top-down BI architecture where power users can explore data and run queries without affecting other users of the system. These so-called sandboxes give power users access to raw or minimally curated data and allow them to upload their own data and combine it with corporate data to conduct analyses. There are different types of sandboxes for each type of power user.

4, Center of Excellence. It’s important for business users, analysts and report developers throughout the organization to engage in a continuous dialogue so they can share ideas and best practices for using data to drive decisions. This sharing network is the essence of a BI Center of Excellence. 

5. Federated BI Organization. To reconcile the two worlds, BI leaders need to build an organizational bridge between top-down and bottom-up worlds of BI. The best way to do this is by creating a federated organization that embeds corporate BI and data professionals in business units to work side by side with business users. By creating a matrixed BI organization, companies get the best of both worlds of BI.

6. Strong leadership. Creating BI sandboxes, a BI center of Excellence and a federated BI organization require strong leadership. BI directors must create a strong vision and roadmap for the organization and gain the backing of corporate executives who are committed to funding and sustaining the BI initiative. Maintaining strong executive support is a full-time job that requires strong sales and marketing skills and a track record of success.

7 Modern Enterprise BI tools. Interestingly, a new generation of enterprise BI tool that is browser-based and multi-tenant and provides top-to-bottom BI functionality on an integrated BI platform, provides adequate support for both worlds of BI. These emerging modern enterprise BI tools will be critical to unifying the two worlds of BI. (See “Ten Characteristics of a Modern Enterprise BI Tool”)


BI leaders make the mistake of treating everyone the same, and thus alienating everyone.


Summary

Many BI deployments fail because BI leaders don’t know their users. They don’t fully recognize the distinction between casual and power users and the two distinct worlds of BI they inhabit. Without this knowledge, BI leaders make the mistake of treating everyone the same, and thus alienating everyone.

To succeed with BI, organizations need to recognize there are two worlds of BI and then build organizational and technical architectures that treat casual and power users differently. Paradoxically, by dividing BI into two camps with unique set of users, architectures and tools, BI leaders can create a unified BI environment that delivers enormous business value.

Wayne Eckerson

Wayne Eckerson is an internationally recognized thought leader in the business intelligence and analytics field. He is a sought-after consultant and noted speaker who thinks critically, writes clearly and presents...

More About Wayne Eckerson