Here now is the broader conceptual model that I promised in the prior blog post. As I said, I built conceptual hooks in my decision support ROI model to address broader requirements for decision automation and decision management.
Fundamentally, the core hook is the notion of a “decision agent,” which may be a person processing information, evaluating options, and taking action, or may be an automated software component doing all the same. Many business processes involve a complex choreography of people and programs making decisions on many levels.
It’s in this latter context–programmatic elements as decision agents–that we speak of “decision automation,” which is enabled through “decision engines” of all shapes and sizes (rules engines, workflow engines, recommendation engines, etc.). Clearly, automation can increase decision velocity by several orders of magnitude, because an automated program can access more information more quickly, pose more queries, evaluate more options, and take more actions than can any mortal human. Whereas we would evaluate the cost of a human decision agent as their fully burdened salary, we would cost an automated decision agent as the fully burdened expenses necessary for its development, integration, and administration.
In most practical circumstances, decision automation is seldom complete, because there’s usually a limit to the range of decisions that programmatic components can make on their own. Decision engines are usually set up to take as many automated decisions as they can in accordance with complex rule bases. They offload the most routine, repetitive, cut-and-dried decisions from human decision agents. But they still must escalate the “exception conditions” to people for manual resolution. Human beings, as “exception handlers,” are still very much in the loop on most automated business processes.
Consequently, we should view decision automation as more a matter of “decision deflection.” Usually, we should aim to reduce the burden of robotic decisions on human decision makers while also improving decision support on those exceptional matters that are escalated to human exception-handlers for resolution. When people are in the loop, we must improve the level of decision support that is provided to them. Process automation environments accomplish this by delivering to human decision makers only those information and options that are most relevant to the exception conditions under scrutiny. High-quality, consolidated, conformed analytic data must be available to all decision agents — human or automated — to support their ongoing collaboration in a broader decision management environment.
The ROI of decision automation is beyond dispute. It can significantly improve the velocity, quality, and consistency of decisions. It can also allow organizations to significantly reduce operational staffing while freeing up staff to focus on higher-value decisions that don’t lend themselves well to automation. We should view decision automation as an approach that is best suited to the most “regular” business scenarios—in other words, those for which there are clear cut policies, procedures, policies, processes, practices, and precedents that can be rendered as crisp rules. By the same token, we should view decision support as most applicable to “exceptional” business circumstances where such rules must be worked out on the fly by human beings.
Exception handling is integral to most operational business processes, but, looked at more broadly, it’s what your CEO and the rest of your management team do every single day. The business world is chock full of exceptional circumstances—opportunities and threats, innovations and challenges—that demand creative responses of a type that only human beings can improvise using all resources at their disposal. In fact, when the business environment has been turned topsy-turvy by new competitors, conditions, and technologies, exceptional conditions become the rule and creative responses must prevail. In these cases, businesses face exceptional challenges and opportunities. They have to potential to reap exceptional payoffs, perhaps at a multiple of the internal rate of return that they realize in more regular times, but they also face exceptional risks.
Hard-and-fast decision automation is a suicide machine in any exceptional environment where the old rules and business models no longer apply. Likewise, when the situation stabilizes into a regular new order, seat-of-the-pants management becomes ruinously wasteful, draining your budget on ad-hoc processes that can and should be automated to the hilt.
Your core strategic imperatives are threefold. First, implement a flexible decision management environment that allows you to combine the best of decision automation and decision support, whatever makes most sense in your competitive environment. Second, automate as many decisions as possible so that all levels of management can only focus on the higher-value exception conditions that make a difference in whether your business succeeds or fails in the competitive fray. Third, and most important, leverage your decision support, collaboration, and knowledge management environments to help your people craft exceptional innovations that drive exceptional growth.
Make your company the exception in a cut-throat business environment. Make your decision-driven business model the new rule and paradigm that all others must follow.
That’s what’s often known as creative destruction.