Finding Value in Analytics, Part II: Our BI Legacy
In this series of articles, our journey is finding the value generated by analytics within enterprise IT systems. In the first article, we set the stage for this journey by introducing the economic value principle:
Information has economic value only if organizations are enabled to make actionable choices that yield higher expected payoffs than if those choices were made in the absence of that information.
The implication is…to have value, analytics should enable the organization to execute improved actions that, in aggregate, modifies organization behavior toward customers and the like. The objective is a positive net value.
In this second blog, we will explore our seven-decade legacy using various technologies to support Business Intelligence. BI is increasingly relying on machine-learning (ML) analytics that is known informally as artificial intelligence (AI) and properly as deep learning using artificial neural networks.
The table below is an overview of those decades in terms of the value generated by the IT infrastructure. Note the general flow of the value definition (right column) through the decades. And, don’t be distracted by the details.
|keypunch, mag tapes, line printer, mainframes, COBOL, FORTRAN
|Reducing internal costs and errors of manual clerical processing
|System-of-Record, merge-sort, OR/MS, ERP
|Increasing efficiency of administrating financial and then other functional areas
|Database Management & Decision Support
|CODASYL, relational normalization, joins, static SQL, IFPS, what-if cases, PC
|Better choices for big decisions by a few executives; plus, efficiency in data mgt
|Online Database Systems
|dynamic SQL, online reports, spreadsheets, dashboards, KPI
|Better choices for small frequent decisions by many employees
|SVOT, ER model, subject areas, ETL-ELT, star schemas, data marts, ODS, Active DW, mobile
|Data integration for Single Version of the Truth enabling cross-functional insights to optimize corporate-wide processes
|Visual Descriptive Analytics
|drill-down, trending, ease-of-use UI, Big Data, insights
|Insights into data patterns that improve policies and procedures by management
|Interactive Predictive Analytics
|Big Data, machine learning,
|Statistical predictions based on historical data for better planning of biz operations
Do you see the flow patterns in these BI periods? Notice the impacts of certain IT technologies, such as the organization of databases, the concept of decision support, cross-functional views of data warehouses, and interactivity of visual analysis.
Here are some patterns to consider:
Cumulative: The first pattern that struck me is that the technology from each of these periods is present, alive, and improved today. Prior technology is not destroyed, but it stimulates, promotes, and blends into the next. We still do report generation, big time! We just do not kill trees to accomplish it. In the 1950s, interactive visual analysis was performed by plotting points manually on the blackboard while someone shouted out the numbers from a stack of green-lined printer output. That was the early beginnings of VisiCalc, 1-2-3, Excel and now Qlik & Tableau.
Pervasiveness: BI emerges explicitly around the 1960s as Decision Support Systems, which focuses on infrequent big decisions by executives. Subsequent periods progressively increased BI pervasiveness throughout the entire organization. Hence, BI today is about smart enterprise systems that change overall organizational behavior, both internally for thousands of employees and externally via web-based services to millions of customers.
Disruption: Each BI period contains the seeds for disruptive innovation, thus enabling the emergence of the next period. The infrastructure and practices of the previous era endure (with modifications) and enable its successor to exceed. As this blending into the new BI period occurs, there are dramatic shifts in concepts and twists in practices, resulting in disruptive infrastructure changes.
Actionable: Each BI period offers a different value sequence for transforming data into action. In the literature, this sequence is described as a value chain where raw materials (data) are transformed into finished goods (actions) via a sequence of processes, the sum of which yields a net economic value.
Generalizing: The earlier periods focused solely on managing and describing known data, while the latter periods increasing emphasized generalizing beyond known data. This is the essence of being analytic. Generalizing is the inclusive term for inferring or predicting. The conceptual leap from describing to generalizing is one that everyone performs naturally and quickly, but they also perform predictably irrationally and sub-optimally. Capabilities of future analytics to generalize effectively will become the major value generator, as analytics teases every gram of actionable information from our big data.
Effectiveness: The earlier periods were justified by reducing current administrative costs. This generates value via efficiency, which implies performing pre-defined actions better for a generic business situation. The latter periods shifted toward generating value via effectiveness, which implies performing better actions tailored to specific business situations. The key factor is the variability of the actions required for the situation. The extremes are a single market of a million customers versus a million markets of a single customer.
Look around your organization for the numerous ways that BI systems are assisting your business. Become aware of this value perspective across your organization. What if pieces of that infrastructure cease to function? Describe how economic value would change? What would be the impacts on the financial statement?
Note how analytics specifically contributes value. Start imagining that Excel (or similar) stops working. Then, imagine more advanced forms of data analysis tools or systems ceasing to work. What is the value generated from each? What are the impacts on the financials?
Where are we heading on this value-finding journey?
What will be the value generators for the 2020 decade? The next blog will offer clues based on the above flow patterns. We will also dive into distinctive implications of next-gen analytics based on neural networks, such as thinking casually.
In summary, BI has given us a rich legacy and an exciting future, assuming that we, as a global BI professional community, can guide its usage over the next decade effectively and properly for everyone’s benefit.