Policy hubs and closed loops: Clockwise vs. counterclockwise market strategies
By Henry Morris
IDC has postulated the need for a policy hub as the critical link from analytics to action in a closed loop system. Information flows in a clockwise direction from operations to tracking of business results to analysis/modeling to decision-making (policy hub), then back to operations. The policy hub is that point in the business process where current rules are evaluated, adjustments are made and revised rules are forwarded to all relevant operational systems.
Policy hub: linking analytics and business operations.
The figure on this page shows the steps in a business process that combine operational and analytical elements. The policy hub is the critical link between business operations and the analysis of business results.
A transactional or operational system such as enterprise resource planning (ERP) covers:
- Adjust/act—Business rules are established or modified governing specific operations, e.g. pricing for order management. That is necessary for processing transactions.;
An analytic application such as customer (or employee) retention covers several or all of the following:
- Track—Results are monitored to get a reading on the state of current operations, e.g. the current marketing or recruitment campaign. Those results can be compared to targets or goals that have been established.;
- Analyze—Time-oriented data from multiple systems is integrated into a data warehouse or mart in order to support an analysis of key trends. Deviations from expected results or targets are explored, such as factors associated with customer (or employee) attrition. ;
- Model—Models are built to predict the impact of variable factors on business results, such as the likely impact on sales by changing prices to certain products, combinations of products, or products for particular groups of customers.;
- Policy hub—It is the decision-making process. The results of the analysis and modeling work are considered along with business judgment and knowledge to decide on changes or adjustments to business policies or rules. Software automating the policy hub should enable decision-makers to compare the impact of alternative policies, track the decisions made, and review how effective they were.;
The policy hub is the critical link between analytics and operations. Revisions to current policies must be forwarded to the relevant operational systems where specific adjustments are made (e.g. pricing changes) that govern current operations (act). There may be several systems that must be adjusted, hence the need for a hub. The results from multiple systems are monitored (track) and analyzed as the cycle continues.
A closed loop business process, therefore, incorporates feedback on the effectiveness of current operations to enable intelligent adjustments in business rules. We learn from the analysis of trends and causal relationships seen in past results so that we can positively impact the future.
Viewing an optimizing business process as a closed loop makes sense in theory. But business processes seldom work that way in practice. The policy hub is frequently the weak link, due to massive dysfunction in the way decisions are made.
Policy hubs in practice: decision-making dysfunction
Research in psychology and decision sciences have challenged the assumption that people apply rational principles in making decisions that work to optimize expected results. That descriptive, empirical research has cataloged many of the biases that enter into decision-making. Managers often seek to mitigate uncertainty by any means before committing to a decision. (An excellent review of the literature can be found in Max Bazerman, "Judgment in Managerial Decision Making," John Wiley & Sons, 1998.)
One of the most famous examples of that line of research is the study of “anchoring” conducted by Professors Kahneman and Tversky in the 1970s. Through many experiments, they were able to show that managers make judgments by making adjustments from some initial value, even if that initial value is based on totally random information.
Eliminating the dysfunction: lessons from knowledge management
The research from social psychology suggests that the state of managerial decision-making leaves much room for improvement. For decision support systems to be effective in optimizing decision-making, they need to:
- Eliminate the disconnect between technical users and decision-makers—Decision support implies a technical user bringing together information for the benefit of a business manager who has responsibility to make decisions. In that scenario, the actual user of the software is not empowered to make decisions. How likely is it that the information provided to the decision-maker will be relevant? ;
- Support collaboration—Decision support and business intelligence have focused on the individual decision maker, yet decision-making is often a collaborative process employing strategies such as negotiation. Decision process automation software should coordinate team collaboration for decision-making. In fact, analytic applications vendors are beginning to incorporate collaborative support. Two examples are e.Intelligence with its collaborative hub for supply chain analytics and MIS AG (mis-ag.de) with its Collaborative Analytic Processing.;
- Track decisions—Decision support implies that the software is complementary to decision-making. In other words, the process of decision-making is outside the domain of a decision support system. Decisions themselves are not tracked by a decision support system. Hence there is no learning about which decisions were effective and which were not. If there is any tracking of decisions in an organization, it is likely to be in the form of project reports that document lessons learned—i.e. in the form of text that is not accessible by traditional business intelligence tools.;
- Capture best practice processes for decision making—The best performers in an organization use information to make better quality decisions, such as a sales rep qualifying leads or a customer service specialist handling difficult cases. Capturing decision-making processes addresses the fundamental goal of knowledge management: preserving employee expertise. ;
Knowledge management emphasizes capturing the intellectual capital of an organization’s best performers in using information to make decisions that advance business goals. That involves structuring a process in which groups of knowledge workers collaborate to arrive at an effective decision. Documenting the process and the lessons learned enables others in similar roles within the organization to learn from experience and become more productive.
Decision process automation and analytic applications
The focus on the decision-making process and policy hub characterizes decision process automation. A defining characteristic of analytic applications is business process support. However, few analytic applications automate that aspect of the business process. Hence, decision process automation is the next logical direction for analytic applications that seek to demonstrate a measurable impact on business operations.
Getting to closed loop: clockwise vs. counterclockwise strategies
The closed loop represents a learning process that incorporates feedback on business performance to drive decisions on adjusting rules for future business operations. Information flows in a clockwise direction from operations to tracking business results to analysis/modeling (decision support) to the policy hub (decision-making), then back to operations.
But how does an organization proceed toward designing and implementing such a system? The alternative approaches are clockwise and counterclockwise:
- Clockwise design and implementation—Begin with operational data and work forward. Design a data model for analysis based on user requirements for information (i.e. traditional decision support), integrate the available data and then build reports and cubes. (That is the heart of traditional decision support or business intelligence.) Moving in a clockwise direction, the integrated data is used for building predictive models, serving as input for the decision-making process.;
- Counterclockwise design and implementation—Begin with the decision-making process and then work backward. Examine the potential adjustments that can be made to operations. Then design a process according to which alternative rules or policies are evaluated using probability analysis, in order to reach an outcome that has the highest expected value. Data and analysis requirements are defined insofar as they directly support that process, some information gathered through traditional decision support means or through expert judgment and other unstructured sources.
Business intelligence software should be enhanced to incorporate decision-making or decision process automation. From a technological perspective, that includes integration with rules engines (process support), collaboration and content management. From a process perspective, it includes automating the decision-making process (policy hub), not just the data gathering and information creation process (track, analyze and model).;
Although the movement from reporting to analysis to modeling to the policy hub shows steady progress, there are fundamental gaps in such strategies that began with data preparation and analysis. None of those efforts are centered on decision-making. Hence, none of the available decision support tools track the actual decisions in order to learn which decisions were effective and to improve the decision-making process.
The last mile: closing the loop via rules delivery
Decision-centric analytic applications attack the problem of how to optimize decision-making processes via a policy hub. The result is the selection of a rule or set of rules that must be delivered to all transactional systems where the rules are operative. That is the hub aspect of the policy hub. Thus, the challenges involved in maintaining a policy hub are twofold:
- Policy evaluation—Managing rules at a central point, coordinating their adjustment based on the latest analysis, comprising the decision-making process.;
- Policy delivery—Distributing the updated rules to all relevant transactional systems.;
The delivery of the rules is the “last mile” or final step in a closed loop. Today rules delivery is a manual process, but integration is proceeding quickly.
Although the closed loop is executed in a clockwise direction, the design of the system should be counterclockwise—i.e. the capacity to adjust operations leads to the design of the decision process that drives the data and analysis requirements. Clockwise design can lead to collecting and analyzing data that is not relevant to making decisions on potential changes to operations.
From a market perspective, most decision support vendors have built their strategies from a clockwise direction, risking a disconnect with the decision process itself. There are signs, however, that some user organizations and vendors are changing that direction, applying counterclockwise strategies to effect decision process automation.
Henry Morris is VP for Applications and Information Access, IDC (idc.com), e-mail firstname.lastname@example.org.