What Makes Information Workers Productive

As the economies of industrialized countries have become more and more knowledge-based, the question of how to quantify, analyze and study the productivity of workers in information-intensive jobs has become much more important. After all, productivity in a steel mill or an automotive manufacturing plant can be measured in terms of tangible, easily quantifiable products — and many of the processes that produce those outputs are comparatively straightforward to observe and study. However, an information worker’s productivity is often considerably more amorphous than that of a steelworker or an assembly-line worker. What’s more, identifying the factors that contribute to productivity in an information-intensive job is complex.

Now, several researchers have developed a new strategy for gaining insights into information-worker productivity — and their research has yielded some intriguing findings about factors that contribute to the effectiveness of knowledge workers. These findings are discussed in two working papers, Information, Technology and Information Worker Productivity: Task Level Evidence (October 2006; given the best paper award at the 2006 International Conference on Information Systems) and Productivity Effects of Information Diffusion in Networks (May 2007), and are part of a multiyear study supported by Cisco Systems Inc., via the MIT Center for Digital Business. The authors are Sinan Aral, assistant professor at the Leonard N. Stern School of Business at New York University; Erik Brynjolfsson, the Schussel Professor of Management at the MIT Sloan School of Management and director of the MIT Center for Digital Business; and Marshall Van Alstyne, associate professor at Boston University School of Management.

To conduct their research, the authors studied a midsized executive recruiting company over a five-year period. In the recruiting company, output could be measured in terms of completed recruiting projects, each representing a certain amount of revenue for the company. The researchers used detailed accounting records from the company that included information about revenue per employee, employee workload and compensation and the time required to complete recruiting projects. The researchers also conducted surveys of employees to learn about factors such as the extent to which the recruiters used information technology tools, such as databases and e-mail.

An important component of the research was access to information from the recruiting company’s e-mail archives. With data encrypted to preserve individuals’ privacy (in other words, the researchers had no knowledge of what words were contained in any of the e-mails, although they could determine how frequently a given encoded word appeared in e-mails), the authors analyzed patterns found in 10 months’ worth of e-mail traffic — more than 125,000 e-mail messages in all. (Spam was excluded by only analyzing e-mail from contacts outside the company if at least one person in the company had sent a message to that external e-mail address.) Their results yield fascinating insights into information-worker productivity.

One of the study’s interesting findings is that, while increased use of information technology by workers was correlated with increased revenue generation, IT use wasn’t correlated with completing projects more quickly. Contrary to theories about contemporary businesses operating at a quicker “Internet speed,” the researchers found that IT use among recruiters was, in fact, actually correlated with their taking longer to complete projects.

What’s behind this seemingly paradoxical finding? Multitasking. It turns out that, at least in this company and this study, information technology use correlated with improved productivity (as measured by revenue generation, in this case) — but it also correlated with employees working on more projects at once. In other words, workers who were heavier users of information technology tools like e-mail tended to generate more revenue — but were not quicker at completing any given project, because they were working on more projects at the same time.

The study also found that, while some multitasking yields productivity benefits, there are distinct limits to those benefits. In fact, the study found that the relationship between output and multitasking formed a curve that was concave, like an upside-down U. In other words, working on more projects in one time period at first increases productivity, but, as the level of multitasking increases, the marginal benefits of additional multitasking decline — and, at a certain point, taking on still more tasks makes workers less productive rather than more so. After examining several possible explanations for these data, the authors conclude that this finding is most likely a reflection of the fact that people can juggle only a finite number of tasks effectively, and that, as other research has shown, there are cognitive costs associated with switching between tasks. Essentially, they suggest that excessive multitasking may result in the workflow equivalent of a traffic jam, where projects get backed up behind other projects much the way cars get stuck in traffic when there are too many on a highway at once.

Another interesting finding concerned the benefits of being what Aral, Brynjolfsson and Van Alstyne call “a communications middleman.” When they analyzed the company’s e-mail traffic, the researchers discovered that employees who were heavy multitaskers and sat at the center of a lot of e-mail flow within the company were more productive on average than their less well-networked colleagues. In particular, having a diverse network of contacts was associated with higher productivity.

Aral, Brynjolfsson and Van Alstyne also studied the way information diffused via e-mail through the recruiting company. They identified two different types of information flows: “event news,” which involved certain words suddenly showing a substantial but brief spike in usage in the company’s e-mail traffic and diffusing quickly through the organization, and “discussion topics,” which were e-mail conversations that went back and forth between a smaller number of parties. The authors discovered that “discussion topics” showed a tendency to move up and down the organizational hierarchy and that there were strong correlations between people having ties to each other — for example, having worked on projects together in the past — and their likelihood of taking part in an e-mail discussion together. News, meanwhile, seemed to spread through the company without much regard to the strength of the connections between individuals, although men were for some unknown reason far more likely (more than 50% more likely) to receive news than women. Demographic differences, such as differences in education or gender, generally made it less likely that a person would receive e-mail information from a colleague. What’s more, receiving information via e-mail quickly was a strong predictor of increased productivity: The researchers found that workers who encountered just 10 more novel words than the average during the period studied generated $700 more in revenue for the company. At least in this study, then, knowledge really is power — or at least money. And, the authors conclude, information does not spread randomly through organizations; instead, the flow of information within a company reflects the way that people’s relationships are structured.

Because this research was limited to one company, in one industry, it’s difficult to say how many of the specific findings would hold true for other information-intensive industries — or even other executive recruiting companies. The authors observe, however, that their research technique holds promise for future management research.

Information, Technology and Information Worker Productivity: Task Level Evidence is currently available for download at http://ssrn.com/abstract=942310, and Productivity Effects of Information Diffusion in Networks is currently available for download at http://ssrn.com/abstract=987499. A related working paper, Network Structure and Information Advantage (July 2007) by Sinan Aral and Marshall Van Alstyne, is also available for download at http://ssrn.com/abstract=958158. For more information, contact Sinan Aral at sinan@stern.nyu.edu, Erik Brynjolfsson at erikb@mit.edu and Marshall Van Alstyne at mva@bu.edu.