jim.shamlin.com

12: The Right Tool for the Right User

The chapter begins with an anecdote from Dow Chemical, which sought to implement a "global reporting" tool in 1994, ahead of vendor solutions for BI, and met with "resistance on all sides" from users who found the tools too difficult to use, departments that clung to the reporting formats they were getting from older systems, etc. and ultimately were unable to get adequate buy-in from anyone outside the IT department and logistics. The example illustrates two fundamental mistakes: the failure to get sufficient input when selecting a solution, and a failure to build consensus along the way.

The Importance of BI Tools

While IT departments focus on the back-end systems that handle the data, the business users focus their attention on the front-end tools that provide the access and functionality they need. No matter how well the back-end is executed, BI will fail if the front-end tools to not suit the needs of the users.

While business users do not entirely ignore the back-end, the focus on their concern is the value it delivers: the back-end must support the user tools by providing accessible, high-quality data. The study findings suggest that business users will tolerate some degree of reliability in exchange for better tools.

The Role of BI Standardization

The value of standardization to vendors is in having a package solution that can be sold to multiple clients; and the value of standardization to IT departments is the ready availability of documentation, support, and the ability to hire pre-trained personnel. However, this tends to lead to the adoption of off-the-rack solutions that ultimately fail because they do not suit the needs of the business.

Internal standardization, however, is of high importance, in that the data available and analyses performed muse be common within an organization, providing users with information that will be commonly accepted across the organization (rather than each department coming to the table with different information). But even that can be taken too far: in that a single set of standard reports and tools will not be useful for all departments: hence users will need access to different dashboards, reports, and queries that suit their needs.

Problems can arise when companies take a multi-vendor approach to their enterprise BI systems: while various vendors have competency in certain areas, and can serve the needs of some components of the business, this comes at a cost of redundancy, duplication, and inconsistency in interpretation, which should be addressed on the enterprise level.

Contrary to the obvious conclusion (developing custom tools that are internally standard), the author's survey found that companies that use a single-vendor solution report the highest level of success with their BI systems than those that custom solutions. It's also noted that 75% of companies switch BI vendors during the course of their deployment, citing various reasons, such as licensing costs, vendor complacency, and the rigidity of the solution.

The Right Tool for the Right User

A common mistake in considering BI is in an attempt to define a standard toolset to be provided to all users, as if they had the same needs. A senior executive and a sales representative have different functions, hence different needs, and it doesn't merely apply to the "level" of data analyzed, but to the analysis itself. A better approach is to segment the user base into groups with similar characteristics, needs, and desired benefits.

The author describes some of the key characteristics that identify a BI user segment:

The Most Successful BI Module

Across the body of survey respondents, there was one tool that was most frequently used by successful BI deployments: Microsoft Excel. Contrary to the level of hype given to "dashboards" and predictive analytics, the most successful BI implementations rely heavily upon ad-hoc query capabilities that provide data tables that can be loaded into Excel for more detailed and unconstrained analysis.

There are, however, some drawbacks to integrating BI systems with Excel. A few problems listed include inaccurate data being fed to spreadsheets for analysis, a simple spreadsheet error can result in incorrect analyses, and enabling employees to download spreadsheet data to laptop computers may be a security risk (if the laptop is stolen and the data is not properly encrypted).

In spite of these problems, respondents to an informal poll at an event indicated that a large percentage of BI users (67%) indicated that more than half of their reports are routinely exported to Excel, and follow-up polls at other events have been largely consistent with this ratio.