Editor's note: Renee Entinghe is a research analyst with Maritz Marketing Research Inc.'s Automotive Research Group in Toledo, Ohio.

In today's complex marketplace, numerous factors affect customer satisfaction with a product or service. As a result, surveys ask people to evaluate a product or service on multiple attributes. Customer satisfaction profiles are developed from these data to show a company how well it is performing. The profiles generally take the form of charts and/or tables, but the magnitude of information they contain can overwhelm the reader, especially if competitor data are included.

Radar charts (also called spider charts) can be an effective way to present the data. Traditional radar charts are especially useful when comparing competitors' positions on multiple attributes. They are simple, concise, and can be read quickly. Analysis of the radar chart is continuous. The attribute scores are connected in a circular fashion making the overall and individual relationships easily recognized. In addition, radar charts can be taken a step further by incorporating statistical significance testing. With these modified charts, significant differences among competitors or subgroups are more easily identified than with traditional charts, such as bar charts.

Traditional bar chart

Suppose Company A is interested in examining new car buyers' satisfaction with the dealership facilities. Assume also that Company A is especially concerned with how its dealerships compare to those representing Company B and Company C. Dealership facility

satisfaction scores, obtained from the National Survey of Early Model New Car Buyers, are plotted in a traditional bar chart in Figure 1. To compare Company A's performance with that of Com-pany B and Company C, the reader examines each cluster of bars separately. Several passes through the chart are required to determine Company A's relative position for dealership facility satisfaction.

Traditional radar chart

Compare the bar chart in Figure I to a traditional radar chart, using the same dealership facility satisfaction scores, in Figure 2. The radar chart is not as cluttered as the bar chart, making it more appealing to the reader. The connected pattern of the radar chart allows the reader to see the relative position of Company A to Company B and Company C quicker than with the bar chart, whether the reader is examining the attributes independently or collectively.

Plotting significance tests

In Figures 1 and 2 Company B appears to have lower satisfaction scores than Company A for all attributes, while Company C and Company A seem to be fairly comparable. However, these charts do not indicate which differences are meaningful. Radar charts can be modified to illustrate significant differences between competitors. By plotting the results of two-sample z-tests and incorporating the confidence level into the radar chart, the reader can determine where significant differences occur between manufacturers.

For example, the modified radar chart in Figure 3 shows differences between Company A and its domestic competitors. Company A is represented by the shaded band, which is a 90% confidence level (see key in Figure 3). Points that lie on the band are not statistically different from Company A. Points inside the band represent attribute scores that are significantly worse, whereas points outside the band represent scores that are significantly better than Company A's.

From the radar chart in Figure 3, the reader can quickly conclude the following:

  • Company B is significantly worse than Company A for every attribute measuring dealership facility, especially availability of model in stock.
  • Company C is rated significantly better than Company A for several of the dealership attributes. Company C has the largest advantage for appearance of service facility, but is disadvantaged for model in stock. Company A and Company C earn comparable ratings for overall facility and availability of brochures.

Other considerations

Because z-scores are plotted on the modified radar chart, care must be taken when interpreting the values on the chart The points plotted reflect the difference between the competitor and Company A; they do not represent the actual satisfaction scores of that competitor. For example, in Figure 3 Company B's largest plotted value is for location. Because this Company B data point is closest to the significance band, it implies that the smallest difference between Company A and Company B occurs for location, not that Company B dealerships received the highest score for this attribute.

Conclusions about significant differences can only be drawn regarding the company being profiled, i.e., Company A. For example, in Figure 3 Company C and Company B are significantly worse than Company A for model in stock. Although it appears from Figure 3 that Company C has a higher satisfaction score for model in stock than Company B, it cannot be determined from the chart if this difference is significant. Providing additional charts in which competitors are compared can resolve this situation.

Conclusions

Radar charts are a simple but effective device for presenting complicated data and plotting significance. They are particularly useful when comparing competitors or subgroups on a series of related attributes. Besides customer satisfaction studies, modified radar charts can be effectively used to show significant differences in image, product, and corporate evaluations.