Using Methods of Intellectual Analysis to Step up Profitability of Network Business

Abstract

The article studies the use of drug-store chain clusterization and determines strategy of work with each of the clusters, which is necessary because key work parameters of drug-store chains, such as selling speed, the number of brands of the company being investigated in the chain, the condition of stores, availability of strong competitors in the chain selling must be verified. Another reason is complexity of company resource distribution to build communications with drug-store chains (representatives’ calls) and products themselves among drug-store chains. To resolve the problem of drug-store clusterization k-means method based on the assessment of the company share and the share of a concrete brand in the drug-store chain was chosen for the research. As metrics for identifying the quality of obtained results ‘silhouette’ was chosen, i.e. the form of a cluster set representation. Its factor is equal to 0.514, which testifies to rather high accuracy of results and possibility to introduce this algorithm into real business practice. By cluster analyzing the multitude of drug-store chains a set of three clusters was identified. On the basis of these results of clusterization and current business requirements for each cluster a number of recommendations were put forward aimed at interaction between the pharmaceutical company and drug-store chains in the aspect of assessing the competition environment, analyzing selling speed and re-filling stocks of the chain, frequency of pharmaceutical representatives’ calls to drug-store chains. The findings of the research allowed us to draw a conclusion that any method of clusterization of drug-store data should be renewed in view of the quality of result interpretation, topicality of initial criteria of clusterization and should be corrected proceeding from business requirements, which arrive from marketer teams and managers on medicine categories and separate brands.