Co-design process
Mode 1: Ask customers
5 Empirical Research Design
5.2 Measure Development
5.2.1 Mass Customization Capabilities and Performance Measures
Whenever possible, we used existing measurement instruments for the core constructs. All scales are presented in Appendix 8.2.1. The scale used to measure RPD was adapted from Zhang et al. (2003), using a combination of items from their volume and mix flexibility scales. As discussed above, while firms may apply different methods to increase the robustness of their process designs, the core objective of RPD is to ensure the required levels of volume and mix flexibility so that the firm can efficiently serve its customers individually.
However, established scales for SSD and CN were not available. We therefore generated them specifically for this study based on a rigorous process that focused on attaining content validity by reviewing relevant literature and consulting with company executives. For the pretest, the definitions of the capabilities and measurement items were examined by five academic experts from different universities who had expertise in mass customization, operations management, innovation management, and marketing. To further enhance the content validity, 15 attendees of an executive MBA program on mass customization at Instituto de Empresa, Madrid (Spain) participated in a Q-sort exercise. The managers acted as judges and were asked to independently sort the 15 measurement items into the three predefined and mutually exclusive measurement scales for the strategic capabilities (Rungtusanatham 1998). However, instead of using Cohen’s kappa as a measure of inter-rater reliability, which is only appropriate when assessing the agreement between two raters, we applied Fleiss’ kappa (Fleiss 1971). The resulting kappa value of 0.63 indicates a “substantial agreement” of the raters in assigning the items to the three capabilities (Landis and Koch 1977, p. 165). The results are presented in Appendix 8.2.2.
The scale for overall mass customization capability (MCC) was adapted and revised from Tu et al. (2001), such that the items adequately reflect the four competitive priorities of mass customization firms—namely, quality, responsiveness, scalability, and costs. The final scales for the three strategic capabilities and the overall mass customization capability were five- point Likert-type scales with 1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, and 5 = strongly agree.
Business performance is a crucial indicator when comparing strategic configurations of firms (Ketchen et al. 1993). During the pretest, however, firms were rather reluctant to disclose their absolute sales and performance figures. We thus relied on the market performance scale proposed by Homburg and Pflesser (2000) to obtain a relative measure of performance while preserving the respondents’ privacy. Furthermore, because the market performance scale is oriented more toward customer success, we included “achieving sales growth” as an additional item to reflect economic success. We deliberately did not include any item relating to profitability (e.g., return on sales). Such a measure would not be meaningful for comparing performance among the companies in our sample because many of the recently founded companies were still experiencing the usual start-up losses. As previously discussed, developing the strategic capabilities for mass customization often requires substantial upfront investments (e.g., in configuration technology or flexible automation systems) that must be recovered in the following years. The performance scale was five-point Likert-type with the anchor points 1 = “much worse relative to main competitors,” to 5 = “much better relative to main competitors.”
5.2.2 Antecedents, Contingency Factors, and Controls
Within our conceptual framework, we have derived different sources of information used to develop product offerings and the presence of a formal revision process as possible antecedents for SSD. For the revision process we included a dummy variable (SSD_REV), where firms were assigned the value 1 if they stated that they have a formal process to revise, trim, or extend their solution space. Concerning the sources of information used to develop the product offerings and improve the solution space during the revision process, we differentiate between (1) opportunity recognition heuristics (SSD_OPP), (2) secondary knowledge sources (SSD_SEC), and (3) information obtained directly from customers (SSD_CUST). The use of opportunity recognition heuristics refers to statements such as “we relied on our experience” and “we translated our own unsatisfied needs into a product offering.” Secondary knowledge sources comprise trend reports, external consultants, analysis of purchase data, and benchmarking of competitors’ assortments. Information can also be obtained directly from customers through personal interviews, focus groups and surveys. Firms were asked to indicate on a five-point scale anchored at 1 = “never” and 5 = “frequently” whether they use
the respective sources. The three variables simply represent the average of the firms’ ratings in each category.
To measure the antecedents of RPD, we relied exclusively on existing scales. The scales for product (PDM) and process modularity (PCM) were adapted from Tu et al. (2004b). For flexible automation (FA), we generated one item for each of the four flexibility dimensions (range-number, range-heterogenity, mobility, and uniformity) of the machine flexibility scale proposed by Koste et al. (2004). Finally, to measure human resources flexibility, we adapted the skill flexibility (SF) scale of Bhattacharya et al. (2005).
With regard to the antecedents of CN, we took a twofold approach. First, we developed three measures that firms were asked to indicate directly in the survey. These were (1) the cumulative financial investment in configuration technology (INVEST), (2) the number of revisions of the configurator (CN_REV) to date, and (3) the diversity of information sources used for the revision (CN_INFO). The potential information sources were market studies on configuration technology, industry benchmarking of other configurators, external consultants, analysis of past configurations, customer feedback, and logfile analysis; the respective source was counted if it was ranked 4 or 5 on a five-point scale with the anchors 1 = “very unimportant” to 5 = “very important.” Second, we analyzed the online configurator of each of the 115 firms in the final sample with regard to specific features critical to effective choice navigation that were derived from Randall et al. (2005) and our literature review in Chapter 4.3. These features are needs-based preference elicitation (NEEDS), a default configuration as starting point (DEFAULT), the option to save the configuration and continue at a later time (SAVE), and the possibility to visually compare two configured products side by side (COMPARE). We also tracked whether the configurators provide a shortcut to the shopping cart without running through the entire configuration process (SHORT), have a help function (HELP), offer explanations on the product attributes (EXPLAIN), and show recommendations based on other customers’ selections (PEER). Furthermore, the richness of the visualization (3D) with the possibility to zoom in on objects, rotate them, or view them from different angles (ZOOM) was assessed. Finally, we checked whether each module is priced separately during the configuration process (MODPRICE) as opposed to overall prices and whether customers can share their creations in social networks (COMMUN). For all features we
included a dummy variable where firms are assigned the value 1 if their configurator has the respective features.
The scales for technological turbulence (TT), market turbulence (MT), and competitive intensity (CI) are based on Jaworski and Kohli (1993), and many other studies have used these scales (e.g., Sethi and Iqbal 2008; Lichtenthaler 2009). Moreover, we controlled for a firm’s experience in the market for mass customized products (AGE) which was measured as the period since the launch of the online mass customization offering. Additionally, firm size may also affect mass customization capability. Larger firms usually have a larger resource base, which enables them to develop the three capabilities simultaneously, while smaller firms instead may focus on one or two of them as a priority due to resource constraints. We therefore also controlled for firm size measured as the logarithm of the average number of full-time equivalents (FTE) employed in the fiscal year 2010. Finally, we differentiated between firms that were founded exclusively with the purpose of mass customization and established companies that run their mass customization business in a separate unit, in addition to their standard business (TYPE). The scales are presented in Appendix 8.2.3.
5.2.3 Descriptive Statistics for Antecedents
Before validating the measures, we performed a descriptive analysis of some of the antecedents of SSD and CN to develop a better understanding of the responding firms’ revision cycles, investment expenditures and customer interface designs. The results are provided in Table 5. It can be seen that 36.6% of the mass customizers have implemented a formal process to revise, trim, or extend their solution space at regular intervals. The average cumulative investment in configuration technology across all respondents since the launch of their online mass customization offering amounts to $114,000. During this time, the configurator was revised 2.8 times on average. While the effectiveness of different configurator features in reducing complexity and creating an enjoyable co-design experience is well founded in the mass customization literature, their implementation in practice seems to be lagging behind. Only 1.8% of firms base their solution searching approach on needs, whereas the overwhelming majority still adheres to parameter-based preference elicitation. Only 33.0% of firms provide a rich 3D visualization, and a mere 34.8% allow their customers to turn/rotate the configured product or zoom in on details, which are important facets of process enjoyment.
Only 42.0% offer their customers the possibility to save their configurations and continue at a later time; in the other cases, the customers must start the tedious configuration process from scratch. An explicit help button or hotline could be found only on 22% of the websites. Most firms (58.9%) rely on individual module pricing, although empirical research has demonstrated that it increases the complexity of using a configurator; customers also tend to select less expensive modules, as individual pricing makes prices more salient (Dellaert and Stremersch 2005). However, mass customizers seem to increasingly recognize the potential of social media. 66.1% of the respondents offer customers the possibility to connect with other customer via social networks such as Facebook or Twitter and share product visualizations for critique and collaboration.
Measure Frequency/
Value
Percentage
Antecedents of SSD
Formal revision process (SSD_REV) 41 36.6%
Antecedents of CN
Avg. cumulative investment in configurator in USD (INVEST) 114,000 -- Avg. no. of revisions of configurator since launch (CN_REV) 2.8 -- Configurator features
Needs-based preference elicitation (NEEDS) 2 1.8% Default configuration (DEFAULT) 94 83.9%
3D view (3D) 37 33.0%
Zoom/turn/rotate (ZOOM) 39 34.8%
Visual side-by-side comparison (COMPARE) 51 45.5% Possibility to save configuration (SAVE) 47 42.0%
Help function (HELP) 35 31.3%
Explanation of product attributes (EXPLAIN) 58 51.8% Shortcut to the shopping cart (SHORT) 65 58.0% Recommendations based on selections of others (PEER) 25 22.3% Module pricing (MODPRICE) 66 58.9% Connection to social media (COMMUN) 74 66.1%
Table 5: Descriptive Statistics for Selected Antecedents of SSD and CN