Finding Value in Analytics, Part IV: Action Distance
The journey of this series is finding the value generated by analytics within enterprise IT systems. Why should companies invest resources in analytics as part of their IT infrastructure?
The previous articles in this series have covered:
1. The Journey sets the stage by introducing the economic value principle:
Information has economic value only if we are enabled to make actionable choices that yield higher expected payoffs than if we made choices in the absence of that information.
To have value, analytics must enable the organization to execute improved actions that, in aggregate, changes its behavior and improves its expected payoffs. Hence, the focus shifts to how analytics transforms data into actions, not just insights.
2. Our BI Legacy explores our seven-decade legacy using various IT technologies to support Business Intelligence (BI). Each decade had a unique value generator, for which we noted six trends in flow patterns from one decade to the next: Cumulative, Pervasiveness, Disruption, Actionable, Generalizing, and Effectiveness.
3. Next-Gen futurizes about the value generators for the 2020-decade, based on six trends observed over the past decades of IT systems evolution. The next decade is challenging with the emergence of neural networks analytics, AnalyticOps to manage analytic systems, continuous learning systems, and intelligent effectiveness customizing each transaction.
This fourth blog takes a back-to-the-future step by examining an old concept – Action Distance – in light of the 2020-decade focus on next-gen analytics. The questions are…
What is the distance (broadly defined) between your data and the actions that are performed based on that data?
Does minimizing action distance enable your company to be more responsive to current data?
Should we minimize Action Distance in all business situations?
What is Action Distance?
Around 2003, I wrote several articles on Minimizing Action Distance in DM Review and TDAN.  The focus is on making data more actionable. Action distance was defined as a measure of the required effort (resources and time) to understand the data and to affect actions based on the information within that data. The following figure highlights the concept as the distance between gauges to levers for a specific business situation.
The analogy used in those articles was a (then modern) aircraft, as shown below.
Instruments (gauges, dials) and controls (switches, levers) are intermixed in the cockpit. Over decades of evolution, the design of an airplane cockpit has systematically placed instruments and controls in order to ‘minimize action distance’ for the pilot. For fun, imagine an aircraft with its sensors located in the rear of the aircraft, where the copilot described the situation to the pilot who managed the controls. This is often how companies are designed and managed.
Action Distance as Value-Time Curve
In these articles, action distance was explained in terms of how business value varied from when a business event occurred to when action was taken in response, as shown below.
This Value-Time Curve shows various stages in the process sequence from data to action, each with its time latency for data, analysis, and decision. The action distance is measured in terms of the time delay from the event to action. In this illustration, this value-time curve decays exponentially from the time that a business event occurs, which in hindsight was overly simplistic. 
Back then, the primary concern was pushing IT infrastructure for timely data collection and integration into the data warehouse so that meaningful data could be displayed on a CRT device ASAP. The context was that a person would be alerted when an unusual condition occurred, comprehend the situation from the display, and take swift action. The human was always in the loop! And, the limiting factor was technology …much like the airplane analogy above.
For example, Walmart boasted back then that its system could capture point-of-sale transactions and insert that data into their data warehouse by the time the customer left the parking lot. The business opportunity enabled by this capability was the ability to snail-mail coupons to customers in the coming week. That was the state-of-the-art in early 2000-decade.
Even back then, I joked that Walmart missed their best opportunity because their customers have left the parking lot, eliciting images of Walmart clerks running after cars waving coupons in their hands.
In hindsight, my joke implied that my value-time curve was wrong. There is a huge value spike at the moment the sale at the cash register was made by the business opportunity to present personalized coupons to customers for future purchases. That value quickly declines as the customer exits the store. At my local store, this is now an automatic standard procedure.
Action Distance in the 2020 Decade of Analytics
In the 2020-decade, enterprise systems are changing rapidly, driven by innovation in analytic systems. Is the concept of action distance still relevant?
The simple answer is yes, but… it is more complicated with next-gen analytics.
In the past, action distance was primarily determined by the resources and time devoted to data capture and integration. That data infrastructure constrained the Value-Time Curve, limiting the timing of business interactions. Further, the action decision is either made by a person viewing a panel of numbers or by an algorithm executing policies pre-defined by a person. Either way, there was a human in the decision loop.
Now, infrastructures are able to deliver data in sub-second times. However, the business situations are more complex with higher volume and greater variety, attributes of the Big Data environments challenging most organizations. At the scale of most large organizations, having people make decisions on simple data is inadequate. Analytics will be a critical element in managing action distances for either replacing the person or augmenting their capabilities.
For Action Distance and Value-Time Curves to be relevant to the business value of next-gen analytics, several critical issues must be addressed.
Insights Are Precious
Traditional insight generation by current BI systems is still critical for persons to explore and understand business complexities. However, the issue is that the missing capability in BI systems is monitoring whether and how an insight actually results in an action by the organization.
Insights are precious and costly, so treat them as valuable corporate assets. Manage your insight-generation by tracking their lifecycle and assigning a tangible value metric to each. Encourage the generation of the most valuable insights that actually result in actions.
Creativity is Demanding
Managers will require new levels of creativity to develop and operate analytic systems since basic concepts and best practices are constantly changing. The definition of a business use case is a critical step requiring imagination that will seem counterintuitive. For example, conceiving of possible Value-Time Curve for a special Action Distance requires thinking creatively about the business situation and its value opportunities across the entire time spectrum, from milliseconds to years.
Trust Is Essential
In the past, trust was derived from the persons who were accountable for decisions that led to actions. They could explain their decisions, account to others that the actions taken were proper, and provide an audit of the process.
Maintaining trust in analytic systems is done by managing the explainability, accountability, and auditability for analytic processes at the transaction level. Your organization should be able to explain how this result generated, why this result is reasonable, and who is accountable for this result. Hence, trust has and should be part of the analytic value strategy. But how is trust quantified relative to the tangible resources and time expended?
Role of Humans in Decision-Making
Defining the role of humans in the decision-making process is a critical issue, one which builds upon the previous trust issue. The current trend is toward eliminating persons from analytic systems, rather than augmenting analytic decision-making by retaining human intuition and judgment.
Business Intelligence started with Decision Support Systems to augment human intellect in the 1960s. It is incredible that the same technology (updated by sixty years) is poised to eliminate human involvement, rather than augment it. Analytic-driven automation–using intelligent bots, autonomous vehicles, and recommendation engines–is becoming the preferred alternative for increased efficiency, accuracy, and reliability.
The following questions require our consideration:
What is the business value of human decisions versus analytics-driven automation?
Does human intuition and judgment have a unique value for companies? Building trust?
How does human intuition and judgment integrate into organizational governance?
What are the best practices for integrating Human-In-The-Loop within analytic systems?
When is it proper for analytic automation to eliminate human involvement? And, when not?
The next article in this Finding Value in Analytics series will probe these questions, along with suggestions for dealing with this issue.
 Richard Hackathorn, Minimizing Action Distance, The Data Administration Newsletter, July 1, 2003. A PDF copy is available here.
 This Value-Time curve is too simplistic in most business situations since it assumes a constant decay. The assumption is: The faster an action is taken in response to an event, the better. In any real situation, determine the actual business value of taking appropriate action at specific points across a spectrum of response times. For instance, the Walmart situation has a huge value spike at the moment the sale was made at the cash register from an opportunity to present coupons to the customer for future purchases. That value quickly declines as the sale is concluded and the customer exits the store.