The BI market today places significant emphasis on discovery, self-service and other related aspects of users’ decision-making behavior. This focus tends to de-emphasize another area of BI functionality that I have long believed to be vitally important to business success: a closed-loop process that spans from information gathering and use through to decision making, action taking, and back again to information.

As far back as 2008, I have called this process the MEDA model of organizational decision making, where the acronym represents Monitor, Evaluate, Decide and Act. The model is based on a higher-level management consulting concept established by Stefan Haeckel in the late 1990s called sense and respond in an adaptive enterprise. The basic premise is that an adaptive organization continuously monitors (senses) its environment for change, evaluates possible reactions, and decides and acts (responds) accordingly. This may sound simple and, indeed, obvious. It is—at some level. However, the closed-loop nature of the concept further prescribes that the outcomes of any action must be further monitored to understand if the action taken was effective, ineffective or even counter-productive.

Figure 1. The MEDA Model

The MEDA model, as shown in figure 1, is applicable to all levels of decision making, from operational to strategic, but a tactical BI example is perhaps most easily understood. In a retail business, a sudden drop in sales is observed (monitored). Why has this occurred? The question is taken up by a business analyst and evaluated using a typical BI tool. The conclusion is that sales incentive plans are not driving the expected sales behavior. Options to address this are discussed by management, leading to a decision and, eventually, action to restructure the incentive plan. The loop is closed if the resulting sales patterns are monitored for the effect of the changes made.

The decision making environment in most organizations today seldom lives up to all aspects of this model. The loop is fragmented between different tools, such as dashboards, spreadsheets, BI discovery tools, collaborative tools, and so on. Some areas are supported only poorly, if at all. For example, action taking may be instigated and tracked only through presentations and e-mails. The loop back from action to monitoring the effect of those actions is often neglected entirely. I contend that such ineffectual implementation contributes in part to the relatively limited success enjoyed by BI in many organizations despite years of investment focus.

YellowFin’s release 7.2, which became generally available last March, provides interesting collaboration-based function that aligns well with the model outlined above. While much of the current industry focus is on the interactive aspects of BI—discovery, visualization and storytelling, for example—a process-oriented focus can be seen emerging here.

This is seen in the concept of “smart tasks” that allow actions to be assigned to people and their progress tracked to completion. These tasks can be initiated by business users directly from a dashboard or report or they can be triggered by some predefined trigger in an automated report or dashboard. This latter aspect maps directly to the monitoring phase of MEDA above. The task itself offers the direct link to action. The action could be to request a deeper analysis in the evaluation phase, to offer options from evaluation to the decision phase, or to initiate specific business actions in the final phase of the MEDA loop. Missing, so far at least, is a function to close the loop back from act to monitor.

Of interest also is that the functionality extends easily beyond the pure business process described above. It allows a business user to direct tasks to IT as well. Such tasks could include requests for changes to existing reports, notifications of data quality issues, and so on. Linked to other collaborative functionality such as discussion boards, polling and timelines, we see the first foundation for the type of workflow and process choreography for decision making that I have previously described in Business unintelligence.

For organizations wishing to formalize the processes around decision making, starting within the scope of a BI tool as described here makes perfect sense. Some organizations with more mature and enterprise-wide collaborative and/or workflow implementation may face a challenge: how to position and potentially integrate the BI-based and enterprise-wide sets of workflow functionality. Nonetheless, for the BI industry as a whole, a new opportunity to create decision making processes by and for ordinary business users is a welcome development.