We investigate these two potentially problematic issues associated with the use of growth as a performance measure using both qualitative and quantitative data. Our analysis included in-depth case histories of five high-tech companies and a two-part survey study.


Sample and Research Design to Test Proposition 1

To study growth patterns we conducted a retrospective case history analysis of five high-tech firms (Eisenhardt, 1989; Yin, 1989). We used a theory-based sampling design (Eisenhardt, 1989) to target manufacturers of tangible high-tech products and accentuate similarities among firms. Our sample included companies manufacturing hand-held and stationary data logging devices, porphyrin compounds, laser beam diagnostic instruments, high speed digital systems, and clinical and life science instruments. The companies had been in business for an average of 19.9 years, with a range of 9 to 25 years.

We used a multiple case design that permitted the analysis of contemporary phenomena in real contexts (Eisenhardt, 1989). Researchers have attributed several strengths to the multiple case design, such as its suitability for studying dynamic processes (Yin, 1989). This method facilitates the construction of novel, testable theories with intimate linkages to empirical data.

Two individuals from each company reported information with respect to growth patterns and perceived key events in the development of their companies. We asked respondents to chart key events in the history of their companies on time lines capturing: (1) changes in product or market, (2) acquisition of physical resources, (3) acquisition of human resources, (4) entry into cooperative relationships, (5) changes in management or management team, (6) external events, (7) and sales performance. We then aggregated events into a single time line for each company and sought to decipher patterns of events related to concurrent or subsequent changes in sales.

We used two respondents from each company to provide evidence for the reliability and temporal stability of recall data. In our final analysis, we used only those events recalled by both respondents.


Sample and Methods to Test Proposition 2

Initial survey. In 1990 an initial sample was selected from the State of Utah records of business incorporation. Questionnaires were mailed to the founders of all firms (n=431) in five types of businesses--manufacturing, retail, wholesale, and business services--that met the following criteria: (1) they incorporated between January 1985 and June 1988; (2) they were in good standing with the state as of February, 1990; and (3) they were not subsidiaries of parent corporations. The data collection procedure included an initial questionnaire, a follow-up postcard and a second mailing of the questionnaire two weeks later. Of the 431 questionnaires mailed, 35 were returned by the post office marked "not deliverable." There were 134 completed questionnaires returned, for an effective response rate of 35%.

Questionnaires requesting performance data to validate the information provided by the founders were mailed independently to individuals listed as vice-presidents in each of the firms. Complete matching firm performance data were returned for 38 of the firms. The initial respondents were 92% male and 96% Caucasian, with 85 percent married. Education levels were above average for Utah with 88% having at least some college and 47% holding at least a bachelors degree. Ages ranged from 25 to 65 with a mean of 43 years. Founders reported working a mean of 53 hours per week and had been intensively involved in a median of 1 previous business startup. Completed questionnaires were returned for 134 manufacturing and service firms. The manufacturing firms (45%) represent both low-technology and high-technology industries, with a variety of products ranging from ceramic pottery, upholstered furniture, and animal kennels to medical devices, electronic instruments, and ion lasers. The service firms include retailers (20%), wholesalers, (14%) and business services, (21%). The firms reported median sales of $100,000 to $500,000 and median sales growth of 15% to 19%. Median earnings, including the salary of the founder, ranged from $50,000 to $100,000. The median number of employees was 6. Although the data came from businesses incorporating between 1985 and 1988, 32 of the 134 companies reported that they were in business before 1985. Nineteen companies had been in business between 6 and 10 years; the remaining 13 companies had been in business from 11 to 25 years.

Follow-up survey. In 1994 follow-up questionnaires were mailed to all of the initial respondents (134 founders, 38 vice-presidents). Of the 172 questionnaires mailed, 28 were returned by the post office marked "not deliverable." To find addresses for these individuals, we went to the telephone directories for the cities in which they were previously located. We found new addresses (either business or home) for all but four of the undelivered questionnaires. After finding the new addresses three more mailings of the questionnaires followed during a four month time period. After four mailings we had collected 66 questionnaires from business founders (48% response rate) and 18 from individuals listed as vice-presidents (47%).

No statistically significant differences (p<.10) existed in 1990 sales or sales growth rates among those responding to the initial survey and those responding to the follow-up survey. Significant differences did not exist in industry representation. Likewise, no statistically significant differences existed in founders’ age, sex, race, marital status, education levels, pre-ownership general management experience and technical experience. However, those who responded to the second survey spent a mean of 9.5 years managing a business they owned prior to the current business compared to 6.8 years for non-respondents--a statistically significant difference (p< .05). Thus, with the exception that respondents to the second questionnaire possessed somewhat more experience as business owners, the follow-up survey is representative of the initial sample.

For the purposes of this analysis, we used the past year’s sales growth. Because sales growth was self-reported we used the secondary respondents to check the reliability of measurement. At time A there were 38 companies with both primary and secondary respondents. The Pearson correlation coefficient between primary and secondary respondents was .65 (n=38). At time B there were 18 companies with both primary and secondary respondents. The Pearson correlation coefficient was .86 for sales growth. In 11 of the 18 companies with both primary and secondary respondents, performance indicators were identical. In the remaining companies there were some discrepancies resulting in less than perfect correspondence between the reported performance of primary and secondary respondents. In five cases there was a one category differential, in two cases there were two points of difference. Overall, this indicates substantial agreement in performance ratings.



Proposition 1 Results

Results of this analysis are displayed in figures 1 through 5. Sales growth patterns were plotted for each of the five companies in the sample. We then superimposed significant events with respect to major environmental changes, changes to the management team, addition or deletion of key personnel, and major additions or changes to the product line. The temporal placement of such events provide significant insights into the antecedent and concurrent variables associated with different growth patterns. The temporal sequencing of events as portrayed by the event analyses provides insights into antecedents and correlates of peaks and valleys in sales growth.

Proposition 2 Results

To test the second proposition, we used hierarchical regression analysis. We wanted to see if any evidence of relationships existed between 1990 sales growth and 1994 sales growth. First of all, do sales growth at one period in time predict sales growth in subsequent time periods? Secondly, is there any evidence of curvilinear relationships between these two variables. To test this proposition we designated sales growth from 1993-94 as the dependent variable. We entered 1990 sales into the equation to provide a control for company size. We then allowed sales growth from 1989-90 to enter into the equation. The results, reported in Table 1, show significant but relatively weak predictive value. We plotted the residuals of that equation as suggested by Billingsley, Croft, Huntsberger, and Watson (1986). The plotted residuals showed no evidence of curvilinear relationships. Although no evidence existed, we further tested the proposition by testing a quadratic regression model. To reduce multicollinearity, deviational scores were computed. Sales growth from 1989-1990 was squared and entered into the equation following the entry of the original sales growth variable. This variable yielded nonsignificant results, no evidence of a curvilinear relationship, and thus, no support for proposition number two.

Curvilinear Regression Analysis
Dependent Variable = Sales Growth from 1993-1994


Variable Name

Beta Coefficient

Significance Level

1990 Sales (Control Variable)



Sales Growth 1989-1990



Sales Growth (squared)

quadratic curvilinear term






Adjusted R-Square





The analysis of growth patterns provided interesting insights into possible antecedents and correlates of growth that can be tested with larger sample sizes in future research. We analyzed the patterns of events as they related to firm growth; the findings allow us to propose several testable propositions.

Past research suggests that multiple founders are preferable. By analyzing the patterns of growth and the antecedent and concurrent events it appears that some founders must go (Willard, Krueger & Feeser, 1992) and it is important to develop a stable management team with complementary competencies. Firms exhibiting strong sustained growth accomplished this objective in two ways. Company A was started by four brothers with complementary abilities. They worked together cohesively and established a strong team early. Company D achieved similar results; however, the initial founding team of seven broke up within the first two years, allowing one founder to gain control of the company. He hired key personnel with the needed competencies and created employment opportunities allowing for the long-term employment of key personnel.

Conversely, Company E started rapidly with three initial founders. During year two, a business manager was employed and given a small equity position in the company. Before the end of third year, interpersonal differences and power struggles resulted in two of the original founders leaving the company. One original founder remained along with the business manager. A substantial sum of money was required to buy out the departing founders. The two remaining owners could not agree on a strategic direction for the company. The company exhibited sporadic performance for several years until the business manager bought the interest of the remaining founder. After gaining control of the company, the controlling business manager developed a clear strategic direction and sales increased dramatically.

The stability or instability of the top management team appears to result in subsequent sales performance. Infighting and power struggles that resulted in buy outs were accompanied by lackluster performance or even decline of the organization. This discussion leads to the following proposition:

Proposition A: Firm performance (sales growth patterns) is enhanced when a stable and cohesive management team exists with complementary competencies. The corollary to this proposition is that the lack of a stable and cohesive team is associated with sporadic growth patterns.

A second observation is that marked upturns in performance were often preceded by new product offerings representing a substantial change from previous product offerings. Perhaps of even greater importance is the observation that there was not one new product offering that did not come as a direct result of the acquisition of key personnel. However, acquiring new personnel did not always result in significant product developments. Time and energy spent in product development by already employed personnel resulted in incremental improvements to products. This suggests that a group or team working together develops frameworks that inhibit the ability to develop quantum changes in products. This discussion leads to our second proposition.

Proposition B: The acquisition of key personnel precedes the development of major new product offerings, which in turn will precede upturns in sales.

Longer term declines in sales were often accompanied by a failure to stay close to customers. Company C was diverted by a large contract and failed to keep track of the core business. We asked who the major customers were, and were told that 80% were researchers at universities and research hospitals. The remaining 20% were industrial users. When we asked what the industrial companies used the product for we were told, "We don’t know…they order it and we send it". Company E told us that they believed their initial success occurred because they were initial customers. After several years of rocky operations they came to realize that as they transitioned from users to manufacturers they failed to keep in close contact with the users. Hence, they drifted away from the market. A major upturn in growth came when the company repositioned itself and developed a new focus on customer needs.

Proposition C: Companies that displayed sustained growth developed systems and mechanisms to stay close to customers.

Sharp decreases in sales were accompanied by the drying up of funding sources or abrupt regulatory changes. During the early 1980s Company E developed equipment to monitor gasoline contamination of ground water. Sales for the product grew rapidly, but in 1992 the EPA stopped requiring the owners of underground tanks to monitor. Thus, a regulatory change eliminated the market for a product overnight.

Proposition D: Markets created through changes in legislation and regulation can also disappear suddenly through the whims of regulators or legislators.

In two of the five companies, founders sought to acquire managerial skills by hiring outside managers. In both cases, conflicts emerged between the founders and the managers, this was accompanied by lackluster performance that improved only after the outside manager left the organization. In a third company, the founder resigned because of health reasons. The company steadily declined after the resignation of the founder.

Proposition E: Hiring external managers may result in conflicts about the company’s direction. Such conflicts appear to associate with lackluster or declining sales performance.

Previous Page | Main Menu | Next Page


1997 Babson College All Rights Reserved
Last Updated 1/15/97 by Geoff Goldman & Dennis Valencia

To sign-up for the Center for Entrepreneurial Studies' publication lists,
please register with the
Entrepreneurship WebTeam.