Linking research and revenue

Editor’s note: Betsy Charles is president of Strategic Power, a research and consulting firm in Germantown, Wis.

Company executives often evaluate research proposals with this question in mind: "Will the research increase our company’s bottom line?" Many have had experience with research that had little value to the company. Although the research measured perceptions, it wasn’t linked to revenues. Thus, they tend to avoid research unless they see its practical value. In fact, most perceive research as a cost, rather than as a means to increase revenues.

The purpose of this article is to show how research can increase a company’s bottom line. We present several approaches that link research to revenues, using data from customer satisfaction or employee attitude surveys to measure how specific improvements would increase a company’s revenues.

1. Linking overall satisfaction to revenue.

This approach requires data on each respondent’s overall satisfaction with the performance of a strategic business unit (SBU) and each respondent’s purchases from the SBU over a certain period of time. The first step is to group respondents by their level of overall satisfaction and then to average the purchases of each group. In the following example, we expect revenues to increase $100 per customer if overall satisfaction increases one level (or $50 per customer if overall satisfaction increases a half level), assuming all else is equal:

2. Linking aspects of satisfaction to revenue.

This approach requires data on each respondent’s level of satisfaction with various aspects of the SBU’s performance and each respondent’s purchases over a certain period of time from the SBU. The first step is to link each aspect of satisfaction with purchases using simple regression analysis. In this analysis, the independent variable is an aspect of satisfaction and the dependent variable is purchases from the SBU. Simple regression analysis measures the amount an aspect of satisfaction accounts for purchases. Each regression coefficient measures the relative (metric) importance of each aspect of satisfaction in increasing purchases, assuming that satisfaction causes purchases. (Since purchases may have some impact on satisfaction, we need two years of data to prove causality.)

In a hypothetical example, we measure satisfaction with the performance of the SBU using the same five-point scale as shown above. Simple regression analysis of "Timely Delivery" against "Purchases" generates a regression coefficient of .50. All else being equal, we expect revenues to increase 50 percent over the same time period if the average customer converts from being "Very Dissatisfied" to being "Very Satisfied" with "Timely Delivery." If satisfaction improves by one level, rather than five levels, we expect the revenues to increase 10 percent, rather than 50 percent.

Some aspects of satisfaction may be highly correlated with one another. For example, customers may perceive that "Timely Delivery" and "Product Availability" are identical. If both aspects of satisfaction have a regression coefficient of .50 with purchases, their combined effect on revenue overlaps. Thus, if customers increase their average satisfaction with both "Timely Delivery" and "Product Availability" by one level, we expect revenues to increase 10 percent, rather than 20 percent.

We can prioritize aspects of satisfaction by their impact on revenues and classify the most important aspects as key drivers of revenues for the SBU. The wise company improves important aspects in which its performance is relatively low. The priority of each aspect for improvement depends upon the aspect’s importance and customer dissatisfaction with the SBU’s performance on the aspect. The relative priority of each aspect for improvement is measured as Importance * (1 -- Performance).

Quadrant analysis visually compares the importance of aspects to the company’s performance on each aspect. In simple terms, if an aspect is important and the company’s performance on the aspect is strong, the company should promote this aspect as a comparative advantage. If an aspect is important and the company’s performance is weak, the company should improve this aspect. If an aspect is unimportant, the company should maintain its current performance, whether its performance on the aspect is strong or weak.

3. Linking aspects of satisfaction to revenue for SBUs.

The two previous examples are only relevant if the customer satisfaction survey collects data on the purchases of each respondent. If the survey does not collect this data, we can still link aspects of satisfaction to revenues if we have data on purchases by groups of respondents. For example, the company may group respondents by their location and know the revenues that are generated by each location over a certain period of time. This approach is relevant for any set of SBUs such as sales regions, branches of a company, or franchises of a chain.

The first step is to group the respondents by their SBU and to average the responses of the respondents in each SBU. As a result, our data are at the group level, rather than at the respondent level. For example, if the company has 50 locations, each with about 1,000 respondents, our sample size is 50 locations, rather than 50,000 respondents.

The next step is to insert the revenues for each SBU in our data and to repeat the preceding analysis at the group level. In this approach we use average group scores, rather than respondent scores, as the independent variable, and SBU revenues, rather than respondent purchases, as the dependent variable. As before, the regression coefficient measures the relative (metric) importance of each aspect of satisfaction for increasing revenues. As a result of this analysis, we can prioritize improvements for each SBU by the expected impact of each aspect on the SBU’s revenues.

We can use our data to compare SBUs in several ways. The quality index provides an overview of the SBU’s performance by aggregating Performance * Importance for each aspect of satisfaction. We can benchmark the performance of each SBU compared to the average performance of the SBUs. Significant differences between the performance of an SBU and average performance indicates the SBU’s strengths or opportunities for improvement.

Figure I demonstrates how quadrant analysis can compare the importance of each aspect to each SBU’s performance (and to average performance) on the aspects. In quadrant analysis, the vertical axis measures the importance of various aspects of satisfaction while the horizontal axis measures SBU performance on these aspects. In this example, italic type indicates the performance of a specific SBU and regular type indicates for average performance of all SBUs being evaluated.

In this example, the most important aspects are "Quality," "Service," and "Delivery," while the SBU performs strongest on "Quality," "Service," and "Location." Since this SBU has a strong performance on the important aspects of "Quality" and "Service," these aspects are the SBU’s comparative advantages. When we compare the positioning of these aspects for the SBU and the average SBU, we note that the SBU performs stronger on "Service" and weaker on "Quality" than the average SBU. The SBU has a weak performance on the important aspect of "Delivery," and an even weaker performance than the average SBU. Thus, improving its "Delivery" would greatly improve the revenues of the SBUs. Since "Location" and "Price" are relatively unimportant aspects, improving these aspects would not substantially improve the SBU’s revenues.

4. Linking employee attitudes to revenue for SBUs

Similarly, we can link employee attitudes to revenue if we have data on the attitudes of the employees working at each SBU and the revenues of each SBU over a certain period of time. Since employee attitudes may impact SBU revenue or SBU revenue may impact employee attitudes, we need two years of this data to prove causality.

The first step is to average employee attitudes for each SBU and then insert the revenues for each SBU in our data. Then we link employee attitudes with SBU revenues using simple regression analysis. In this analysis, the independent variable is an employee attitude and the dependent variable is the SBU revenues. The regression coefficient measures the relative (metric) importance of each employee attitude on SBU revenues. As a result of this analysis, we can prioritize improvements in employee attitudes by their expected impact on the revenues of each SBU.

We can compare SBUs overall with the quality index and compare an SBU’s relative specific employee attitudes with benchmarking. For example, significant differences between the SBU’s performance and average performance benchmark the SBU’s strengths or opportunities for improvement. As before, quadrant analysis can compare the importance of each aspect relative to each SBU’s performance (and to average performance).

If the data is available, we can repeat this analysis for employees in each functional specialty. One company, a leader in its industry, used this approach to link employee attitudes and customer satisfaction to revenues. For example, Figures II and III show how the attitudes of its marketing staff affect and are affected by revenues and overall satisfaction. Since the results are confidential, the letters in the figures refer to specific employee attitudes.

Maximize impact

These approaches measure the impact of specific improvements on overall satisfaction and revenues. Any company can use the first approach to linking research to revenues, while the other approaches require data on a set of SBUs. The cost of these approaches is minimal if a company has already collected the data. All approaches are designed to maximize the impact of specific improvements on company revenues.