Data, AI, & Machine Learning

Five years ago, analytics practitioners spent about 95% of their time reporting on the past and only about 5% on analysis. Most companies were not really focused on the issue of analytics. While reporting is good, it's not enough. Companies need to understand why those data turned out the way they did, what it might do in the future and how they might optimize a particular one of those variables. Thomas H. Davenport, President's Chair in Information Technology and Management at Babson College, says the potential for analytics to become a critical tie to decision making remains an untapped opportunity for most companies. In a new MIT SMR interview, Davenport explores why the proliferation of data has not led to better decision-making. Davenport looks at the challenges that lie in the way of analytics being embraced by executives and how they can use analysis to understand and manage their business more effectively.

In his book Managing to Learn, John Shook deconstructs the problem-solving journey of one manager and his mentor, and the management mechanism that guided them. The backstory? Shook knows the journey firsthand.

Assessment tools for better-informing decisions have proliferated. Which ones work?

Creative insights require time "Ò and a pace at odds with today’s accelerated economy.

Avoiding the pitfalls in leveraging customer data.

INTELLIGENCE: RESEARCH BRIEF: Assessing the role of electronic product-recommendation systems.

Firms can derive significant benefits from consciously, proactively, and aggressively managing their explicit and explicable knowledge, which many consider the most important factor of production in the knowledge economy. Doing this in a coherent manner requires aligning a firm's organizational and technical resources and capabilities with its knowledge strategy. However, appropriately explicating tacit knowledge so it can be efficiently and meaningfully shared and reapplied -- especially outside the originating community -- is one of the least understood aspects of knowledge management. This suggests a more fundamental challenge, namely, determining which knowledge an organization should make explicit and which it should leave tacit -- a balance that can affect competitive performance. The management of explicit knowledge utilizes four primary resources that the author details: repositories of explicit knowledge; refineries for accumulating, refining, managing, and distributing the knowledge; organization roles to execute and manage the refining process; and information technologies to support the repositories and processes. On the basis of this concept of knowledge management architecture, a firm can segment knowledge processing into two broad classes: integrative and interactive -- each addressing different knowledge management objectives. Together, these approaches provide a broad set of knowledge-processing capabilities. They support well-structured repositories for managing explicit knowledge, while enabling interaction to integrate tacit knowledge. The author presents two case studies of managing explicit knowledge. One is an example of an integrative architecture for the electronic publishing of knowledge gleaned by industry research analysts. The second illustrates the effective use of an interactive architecture for discussion forums to support servicing customers. Zack also discusses several key issues about the broader organizational context for knowledge management, the design and management of knowledge-processing applications, and the benefits that must accrue to be successful. In summary, organizations that are managing knowledge effectively (1) understand their strategic knowledge requirements, (2) devise a knowledge strategy appropriate to their business strategy, and (3) implement an organizational and technical architecture appropriate to the firm's knowledge-processing needs.

Almost every activity in which the enterprise engages, from the most mundane operation to the most far-reaching decision, requires data. Yet data are rarely managed well: few enterprises know what data they have; people cannot access or use data; and data quality is often low. Furthermore, individuals and business units often hoard data, leading to political battles over ownership. To help enterprises manage data as a business resource, the authors survey the fundamental properties of data and explore the special challenges and opportunities involved in working with data. "Data" consist of "data models," which are the organization's definitions of entities, their attributes, and the relationships among them, and "data values," which are the specific realizations of an attribute of the data model for particular entities. "Data records" are the physical manifestations of data stored in paper files, spreadsheets, and databases. Data have many distinctive qualities: for example, they are intangible; easy to copy, share, and transport; can be destroyed or lost inadvertently; are used for a variety of purposes; and are renewable. These and other properties of data have management implications in five key areas: making arrangements to supply the needs of data users; ensuring that individuals can access data; protecting data security; improving and maintaining data quality; and employing data effectively in operations and strategy. The authors offer a set of prescriptions to help enterprises meet these challenges: institute data quality and data supplier management programs, hone data needs, identify and manage critical information chains, recognize the proper role of technology, develop and disseminate an inventory of data resources, specify the terms under which data can be shared, avoid futile political battles, delineate management accountabilities, and ensure that senior executives lead the data management program.

Errors in data can cost a company millions of dollars, alienate customers, and make implementing new strategies difficult or impossible. The author describes a process AT&;T uses to recognize poor data and improve their quality. He proposes a three-step method for identifying data-quality problems, treating data as an asset, and applying quality systems to the processes that create data.