Over the past several years, organizations have increasingly adopted corporate performance management (CPM) applications to leverage investments in data warehouses and business intelligence (BI) tools as well as comply with new regulations such as Sarbanes-Oxley. Vendors both large and small have rushed to fill demand for these finance-centric applications and are now developing similar ones for use in other business functions, such as sales. 数据挖掘研究院
This has led many pundits to coin a new class of analytics, "operational analytics," or what some have termed "business performance management 2.0." Regardless of its name, this class of application continues the long-promised democratization of BI, where insights into business performance are not just the domain of knowledge workers and analysts, but also front-line customer- and market-facing employees.
Whereas in the past insight focused on customers, operational analytics lends insight into the performance of internal business functions, often designed for employees at the forefront of company operations. The decentralization of decision-making enables an organizational agility unheard of in the past, yet almost assuredly increases the complexity of both business processes and the IT infrastructure.
The root of this complexity lies with newly found decision-making capabilities afforded employees, who can identify process changes that may require alterations to hard-coded operational systems (for example, the manner by which a new claim is handled or the offers presented to a segment of customers). In addition, if the organization is employing the latest customer-centric approaches to doing business, such as offering multiple interaction channels, mass- customized products or services, or segmented marketing and offer management, the number and frequency of potential "decision changes" increases exponentially.
The resultant "decision complexity" begs for a centralized and automated approach to manage the rules and analytics underpinning enterprise systems, otherwise known as enterprise decision management (EDM). By pairing an operational analytics project with an EDM solution, organizations can ensure their investment yields the value they expect. 数据挖掘实验室
Treating Decisions Like Data
EDM solutions separate decisions from enterprise systems, a concept not dissimilar to that of master data management (MDM). It's common knowledge that the odds are against most data warehousing projects succeeding. The sheer time it takes to develop a warehouse, the proliferation of quick-fix data marts created in the interim, and the inability to implement changes quickly have "deep sixed" many a data warehouse project. Similar maladies can affect the expected return on an operational analytics project in the absence of an EDM solution.
At the root of many data warehousing problems is the lack of a centrally managed and shared data "master" defining the enterprise's customers, channels, suppliers, products and other core assets. MDM solutions address this problem, helping to deliver the timely insight promised by data warehouses. Similarly, EDM centrally manages the decisions underlying an organization's processes - a kind of "decision master" -- unleashing the power of operational analytics to truly impact business performance.
Operational Analytics' Impact on Decisions
While not all decisions can be automated, those that can are often subjected to frequent changes due to the conclusions that employees derive based on operational analytics.
For example, a manager in a volume business may identify inefficiencies in the way new orders are processed and request a change to the fulfillment system. But what about the ordering system used by the salespeople? If a similar change is not implemented in the ordering system, it could "break," or a process step no longer required could prevent users from submitting new orders, causing a backlog.
Complexity further escalates if this is a multichannel business and orders can be made directly by customers online, or through a contact center. Marketing automation may also be impacted if offers are based in part on delivery times. And for finance to invoice properly, their systems also need to be aware of the change to maintain the correct day's sales outstanding and accurately calculate cashflow. 数据挖掘研究院
This simple case illustrates how what appears to be a small decision made by one person can impact multiple decisions that are both manual as well as automated. For some highly repetitive decisions such as insurance underwriting, often 95 percent to 99 percent can be automated. EDM has the additional benefit of exposing the rules defining these processes so that other enterprise systems can rapidly recognize and implement them, ensuring a level of consistency not possible without EDM. 数据挖掘研究院
Maximizing the Value of Operational Analytics
By its very nature, operational analytics' ultimate goal is to measure business performance at the operational level and serve as an enabler of change. Unfortunately, traditional BI approaches designed to inform lack the ability to see a decision through to its ultimate destination - an operational system. Even when performed in real time, this approach is optimized for analysis, not action, and lacks both the scale and enterprise-wide scope that agile organizations require.
To span the "insight to action" gap and realize the true potential of operational analytics, EDM solutions can arbitrate business processes involving employees and enterprise systems. For example, the rules defining the aforementioned fulfillment process can be automated, shared by multiple systems and exposed to employees. The speed by which decisions can be executed and changed enables the operational analytics user to implement and measure the impact of process changes. Precision is a key tenet in businesses where very fine changes to rules and processes can yield significant benefits, and EDM solutions enable even small decisions to be isolated and measured.
EDM Solution Components
In the absence of an EDM solution, the decisions underpinning a business are either performed manually, embedded in operational applications or don't happen at all. Decisions embedded in legacy systems, third-party applications or those managed by new business process management (BPM) solutions often lack the scale, flexibility and analytic components provided in EDM solutions. 数据挖掘研究院
Scale: As organizations grow and become more customer-centric, they inherently become more complex. Operational analytics can often be a principle driver of this change. Dedicated EDM solutions offer high- performance decisioning today and can accommodate the growth in decision complexity stemming from business change. 数据挖掘实验室
Flexibility: By maintaining decisions independent of applications and processes, they can be shared among other enterprise applications. This not only allows sophisticated business strategies impacting multiple functions to be implemented quickly, but also ensures that changes can be made just as fast, often due to the insights generated by operational analytics. 数据挖掘研究院
Analytic components: EDM solutions also include advanced analytic capabilities for prediction and optimization. It is often not enough to automate a process with predetermined rules; one must assess the performance of said process and use analytics to predict better outcomes. Furthermore, there are often several actions from which to choose based on predicted outcomes, and EDM solutions include optimization analytics to determine the best approach given business objectives. These analytical technologies are highly complementary to operational analytic applications. 数据挖掘研究院
As an enabler of organizational change, operational analytics can bring newfound agility to an organization - or become another example of a failed project. EDM solutions offer capabilities that can help operational analytics projects reach their potential. 数据挖掘研究院
Aligning Operational Analytics with EDM
Just as not all decisions are suitable for automation, not all operational analytic applications can be readily aligned with high-volume business processes. To understand which applications align with decisions suitable for automation and management - and thus EDM - a new evaluation metric called Decision Yield can be employed. According to the book, Smart Enough Systems, "Decision Yield is a broad-based evaluation metric that reveals the quality of your current decisions and decision processes, and helps you plan, justify and measure improvements to these decision processes."1 Taylor and Raden describe five dimensions of decision effectiveness that underpins a Decision Yield assessment: 数据挖掘研究院
- Precision: How optimal is the decision?
- Consistency: How consistent across divisions and channels and time is the decision?
- Agility: How quickly can you effectively change the way the decision is made when you need to?
- Speed: How quickly can you make the decision?
- Cost: How much does it cost you to make the decision?
To show how Decision Yield can be effective in identifying decisions that can benefit from an EDM approach - and be potentially aligned with operational analytics - the following chart describes the questions posed as part of an assessment.
To measure Decision Yield effectively, Taylor and Raden suggest developing a set of questions that are industry and decision-area specific. For instance, when establishing the Decision Yield for an underwriting decision, questions in the Precision dimension might be "How many tiers do you use in rating risk?" or "How accurately do you predict the cost of claims for new customers?" A retailer may similarly be asked, "How many segments do you use in developing targeted offers?" or a financial services firm asked, "How is your customer service function aligned with customers of varying lifetime value?" These more specific questions drill into the precision, consistency, agility, speed and cost of the actual decision targeted for improvement.
After gathering answers to these questions, a measure of the current state of a decision in each of the five dimensions can be derived. By comparing these results with operational analytic applications in use or those planned, an organization can then see which decision areas could be influenced by users and plan accordingly. The alignment of operational analytic applications with EDM presents the opportunity for closed-loop insight into the effectiveness of automated decisions. 数据挖掘实验室
EDM solutions are a logical and often necessary complement to operational analytics. By empowering their front-line workforce with new insights and measurable objectives, adopters of operational analytics create new opportunities for growth and efficiencies but also risk opening a Pandora's box of complexity that begs for an EDM approach to managing decisions. With the aid of new measurement tools such as Decision Yield, organizations can identify the decisions most in need of automation and ensure that operational analytics deliver the value they expect.
References:
- James Taylor and Neil Raden. Smart Enough Systems. Upper Saddle,
New Jersey : Prentice Hall, 2007.
