9: Relevance
Given the pervasiveness of computerized information systems, it is possible to collect all manner of data, and a data warehouse may be loaded with a massive amount of data, but little of it is meaningful. Beyond the cost of maintaining and storing information, a core problem with BI is in information overload: meaningful information is lost in a sea of useless data. Hence, one of the core functions of BI is to find information that is relevant, in the sense that it has the potential to make a positive impact on meaningful decisions.
Case Studies
The author returns to the case of 1-800-CONTACTS, cited earlier as a company that had a high turnover rate among its agents, and exit interviews identified that the core cause was that job performance measures were tied to things that the agents were incapable of monitoring or controlling for lack of information. While reports were provided, they held information of a general nature (too broad, little actionable material) and not frequently enough to enable them to take corrective action. The BI team evaluated the performance metrics and determined a solution that would enable them to monitor relevant data through an interface that was refreshed every 15 minutes, enabling them to take action. The result was a measurable lift in sales the first week the new dashboard was rolled out.
Emergency Medical Associates, a firm that seeks to predict the spread of disease of incidence of biological attacks, routinely gathered information from emergency treatment centers to analyze similarities in symptoms and complaints, and discovered that the data it was collecting would support statistical modeling that could be used to improve the efficiency of resource management, decrease ER wait times, better manage staffing levels, and enable EMTs to route patients to the facility most capable of dealing with specific conditions.
To improve customer loyalty, Continental Airlines established and incentive plan that included offering complementary upgrades to frequent fliers, though the value of this program was undermined when customers often noticed unfilled seats in business class. The problem was that gate agents were not aware that a seat was available until a few minutes prior to departure, and had many other concerns, so offering "free" upgrades was not a priority for them. By upgrading their information systems and offering compensation incentives (a quarterly bonus for filling the business class cabin), Continental was able to improve the performance of the program and reduce customer attrition.
(EN: Each case study mentions relevant and timely information, but lacks granularity. I expect this is because BI is a competitive advantage, and specific details as to what information is managed is a sensitive matter for the companies named.)
The Role of Incentives
While BI systems have the potential to improve performance, a key factor in their success is that employees utilize them as intended. Too often, a program suffers from the "better mousetrap" fallacy and participation is taken for granted - when in reality, individuals tend to act in ways that have the greatest personal benefit. The example given is of a real estate agent, who may not seek the best price for a given house (which is his fiduciary responsibility to the seller) if accepting a lower price will satisfy other metrics (immediate commission, meeting a monthly sales goal, etc.)
Financial incentives are one way to encourage employee participation, but there are other forms of incentive. An employee may be motivated to provide better service to customers, to be recognized for their competence, to outperform their colleagues, or to have a sense that their job is significantly easier than it was before.
(EN: These merit careful consideration, as not all employees have the same motivation, and some measures may motivate some while demotivating others. For example, a competition may motivate some employees to "win" but it also means that other employees will "lose" and become discouraged.)
Of the options available, the author cites that specific and special incentives for short-term performance have less effectiveness overall than tying metrics to existing incentives and long-term metrics.
Personalization vs. Access Restriction
To be effective, BI systems must be able to personalize data, to provide each user access to data an capabilities that is relevant to each individual's role and performance. However, this is too often applied as "security" restrictions, often for arbitrary reasons, that can be debilitating. For example, a sales manager in one region is forbidden to see data that pertains to other regions, and the inability to consider the information in the context of the rest of the organization deprives the user of potentially valuable insight. There are a limited number of valid reasons for which access to information should be restricted (privacy, confidentiality, and legal restrictions).
Requirements-Driven BI
Within the IT organization, the current model for systems development is driven by user requirements, but successful BI deployments do not follow this model: the users may be aware of a problem, but it requires the insight of a specialist to define a solution, and due to the lack of information currently available to users, they may not have the requisite information to recognize a problem or an opportunity. To be effective, BI cannot be restricted to building only the capabilities that users ask for, but neither should it be taken to the opposite extreme of ignoring the user and designing BI based on data alone.
In many instances, the most effective approach is for a BI expert to study the activities of potential users and apply insight of the available data to determine where relevancies exist and define a solution that, while driven by the goal of improving performance, is not restricted to the limitations of user-defined requirements.