Connector Segmentation

...

IV. How has this analysis helped you to segment the market for ConneCtor?

Segmentation analysis is the basic step of breaking the market into a suitable number of clusters. Further profiling and the correct targeting of the clusters have to be enhanced by discriminant data.

With the correct steps of analysis, we have decided to target Cluster 1 (sales personnel) based on their needs, segment size and growth and other factors. The analysis also indicated that Netlink should set the price with careful investigation since the sales personnel are quite price sensitive. Besides, only with both segmentation and discriminant analysis, could we identify underlying opportunities of doing business with the employers of construction or emergency workers through a better understanding of the organizational decision process. This would provide some development direction for ConneCtor if Netlink hopes to attract the workers (or their employers). For example, Netlink may have to develop a separate model that is easy to see, dust resistant, comes with a more durable case etc. so as to fit in the working environment.

V. Advantages of the used approach

First, it is better to use quantitative method to make rational decision. Given a large population, it is hard to well-define segments by experience. However, with the aid of statistical tools, we can better identify segments after collecting samples. Software is also able to give a clearer picture of cluster number, membership and characteristic.

Second, data with different base variables can be automatically standardized by factor analysis. Questions in a questionnaire might not be have the same variables. Standardizing data can make sure it is comparable and without arbitrary.

Third, discriminant analysis can help identify descriptors for easy reaching and serving the target segment. After identifying the way to differentiate the clusters with discriminant functions of the sample, it helps to further predict the characteristics of the customers in the market.

VI. Concerns of the approach used (data collection, analysis, etc.)

First, there is sampling error. Although there are 160 surveys conducted, data collected might not be representative enough for the whole population and the sample size might not be enough. Also, the numbers of respondent in each occupation are not the same (such as 48 respondents working in sales but 6 in emergency). The huge difference between industries might affect the data validity and reliability. Besides, the distribution of the survey might not be fair. Interviewees may also fake data to make themselves look good or to please the interviewer.

Second, wrong choice of choosing how to pre-process data. One can choose using raw data, standardize data or combines related variables through factor analysis. However, it might be difficult to decide if it has correlated variables with different measuring scale.

Third, the discriminant data might not predict the clusters correctly. In this case, the hit rate is 71.88%, so nearly 30% of cases are not correctly classified.

Fourth, the selection of the number of segments might not be correct. After conducting clustering analysis by either hierarchical clustering or partitioning methods, numeric criteria and managerial judgment are used for selecting the number of cluster. This might not be 100% correct. Even with discrimination data, it is still difficult to have a well-defined, clear cut and reachable segment as they might be overlapping and fuzzy.

Fifth, given the difficulty in defining clusters, it is hard to identify descriptive name for each cluster as the correlations between discriminant functions and each descriptor variable might not be so obvious. Sixth, some questions are not specific and relevant enough. For example, the readership of magazines could not well analyzed an individual as it might not have direct relationship between the type of magazines and the media consumption. Types of magazines are also limited.

VII. Recommendations

First, besides hierarchical clustering, partitioning methods such as k-means could be used for double checking the number of clusters. If the solution of k-means is similar to the solution based on hierarchical clustering, we could increase the accuracy and minimize the risk of having wrong number of clusters at the very beginning of our analysis.

Second, Netlink should use GE analysis to help prioritize target segments. GE analyzes a company on the two bases of industry attractiveness and business strength with a series of weighted factors of different importance (e.g. market competition, channel access, compatibility with different gadgets, segment size, profitability). For instance, segment size and profitability may be allocated a heavier weighting if Netlink aims to maximize the number of profitable sales. The decision could be strategically in line with company objectives.

Third, after selecting which segment to target for our marketing campaign in question 2, we should develop attractive campaign specifically to the designated cluster, i.e. cluster 1. For example, we can put more emphasis on the cell and PIM function when developing our promotional materials as these are the key things cluster 1 concern the most.

Fourth, ConneCtor should keep improving its features. Netlink should invest more on its R&D department to keep developing attractive features accommodating the needs of its potential customers. For example, the mean of monitor is 4.79, which is one of the highest attributes that respondents concern. It reflects the need of having a better display for viewing information. Netlink should thus develop a color, large-sized and high-resolution display at a reasonable price. Some other features like updating keypad, improving the security, increasing its RAM, adding camera and redesigning its appearance can also be considered to increase its competitiveness in the PDA market. Netlink may also develop another model, for example, with better dust resistance and durable case, in order to sell it to construction and emergency workers or to their employers.