The folks who study human behavior for a living say that one of the ways we make sense of our world is by constructing stories. Whether it's something that happens to us or something we see on the news, some of us structure the events as a story to organize our knowledge of the world and present it to others.

According to a recent study, product managers use the same approach to make sense of their scanner data.

The study, "Product Managers' Use of Scanner Data: A Story of Organizational Leaning," is available as a working paper from the Marketing Science Institute (MSI).
 
MSI is a non-profit organization, based in Cambridge, Mass., devoted to investigating marketing related issues.

The report's author, David K. Goldstein, is assistant professor of business administration at Boston University. As part of a larger study of the use of scanner data at five grocery manufacturers, Goldstein interviewed six product managers and two information support personnel at a $200 million subsidiary of a major grocery manufacturer, which is given the fictitious name Butler Foods in the report. He interviewed the Butler employees six months after the company bought scanner data for the first time and three months after it had begun using the data companywide.

Goldstein sought to answer four questions:

1. How are product managers using scanner data?

2. How does the analysis of the data help them learn about changes in their marketing environment?

3. How does this learning affect their decisions on pricing, promotion design and new product evaluation?

4. How does the data affect their relationship with customers and fellow employees (salespeople)?

To learn how each subject worked with the data, Goldstein collected archival data from them which he used to construct a database describing how each subject used the scanner data. He then searched for patterns by comparing that information to what he learned in the individual interviews.

Four-step process

Goldstein netted two main findings. First, product managers complete a four-step process to analyze scanner data, change their stories and share learning with others. Second, managers analyze small subsets of their scanner data. These analyses are guided by experience, which is organized as a series of stories.

Here is Goldstein's description of how one product manager worked on a set of data using the four-step process: "First, a manager examines the [scanner] data tool he used was a scanner data analysis, comparing sales of Butler and competitor products of this variety in the South and in other regions. The manager found that Southerners consumed about as [much] of the competitor's light variety as did people in other regions.

"The product manager then rede-fined the problem as one of taste. He came up with a hunch of his own: Butler's light variety does not taste as good as its competition to Southern consumers. The manager tested this hunch using a new tool. He hired a market research firm to conduct blind taste tests. These tests showed no differences between Butler and its competitors in flavor.

"The product manager finally came up with a third hypothesis: that the South regional sales force was doing a poor job of convincing supermarkets to stock Butler light variety. The scanner data supported this hypothesis, showing that sales of light were relatively poor in the South. He accepted this hypothesis because he saw no plausible alternative. Through this sense-making process the manager modified his stories about regional differences in product sales, about the taste of his product, and about the effectiveness of the South regional sales force in introducing new varieties."

Little analysis

In general, Goldstein found that the product managers at Butler did very little statistical analysis to understand the data, even though they had the proper training and computer software to do so.
Because time constraints are one of the main reasons for this, Goldstein suggests that computers may hold the key to more effective analysis.

"Easy-to-use computer-based tools - possibly built using executive support systems software - would make it easier for manager to spot a decline in sales at a national level and 'drill down' to find the regions or products that might be causing the decline. Alternatively, computer-based intelligent agents or insight generators could scan the marketing database to identify marketing problems or opportunities. . . . Since product managers often do not have the time to carry out a market-level weekly or monthly analysis of the data if no major problems are apparent at the national level, these agents can help them find an important local trend that they might otherwise have ignored until it affected national performance. . . ."

Stories may be answer

How to persuade managers to do more sophisticated analyses? Stories may be the answer, Goldstein says. Since product managers structure their knowledge of their use of analytic tools as a set of stories, if they are going to use more advanced techniques, they have to develop additional stories about the use and benefits of more sophisticated statistical techniques.

"To accomplish this, managers must first be trained in how to use these tools to analyze scanner data," Goldstein writes. "The abstract lessons that they learned in statistics classes must be applied to this real-world data set. For example, they must learn how to use regression to examine the impact of price on sales. Stories about the use and benefits of these techniques must be created. A few experienced product managers could work with statistical experts on relevant problems. If these teams succeeded in discovering new truths, and if their success led to measurable improvements in product performance, stories about the use of statistical models and their benefits could spread to other product managers and the use of statistical models would increase."

Of course, there's no substitute for experience, and Goldstein's research found that without the product manager's experience, the scanner data is just numbers. It's up to the user to put it all together.

 

The report, "Product Managers' Use of Scanner Data: A Story of Organizational Leaning" (report number 93-109) is for sale by MSI. Call 617-491 -2060 for more information.