Research On Product Dynamic Competitive Market Structure From The Perspective Of Behavioral Stream

With the advent of the era of consumption,the current market environment competition is extremely fierce,enterprises struggle to earning a higher competitive advantage for occupying a larger market,while emerging enterprises are often struggling with survival problems;which is largely due to the growing differentiation and personalization of consumers and their consumption ability.Therefore,enterprise managers are concerned about how to build their own brand,improve their product competitiveness and meet the heterogeneous needs of consumers,which is closely related to product development,product positioning and product marketing.The traditional management strategy of working on their own is no longer effective,that is,the market is only decided by developers,which means consumers' no active choices.However,the era of consumer centered business has come,and the products that the enterprises developed on their own will no longer have no advantages.Fortunately,the availability of clickstream data provides an opportunity for this and has also gained the attention of scholars and practitioners.However,scholars mostly focus on the page-level analysis of products,and often ignore the potential information of product-level content and behavior.Therefore,this paper defines the data of consumer's operation behavior on the product in click-stream data as behavioral stream data,and further explores the impact of consumer behavior and the content of product page.The relationship between product market and consumer is connected together,that is,the relationship between supply side and demand side.As mentioned above,the development,positioning and marketing of products are not only determined by the supply side,but the insight of consumer behavior is the premise for attracting consumers and resisting risk of consumer churn.The process of consumer demand realization is described into stage decision-making process,that is,in the product set(universal set),the products that consumers realize,search and browse constitute the awareness set,while the products that consumers have further consideration to purchase constitute their consideration set,the selection set is composed of the purchase alternatives further in the consideration set,and finally the purchase is completed.It can be seen that the concept of consideration set is widely used for its definition that products have a greater possibility of selection.However,the demand of consumers changes in real time,that is,the consideration set of consumers is also dynamic,so it is a challenge to capture the real dynamic needs of consumers.This paper focuses on the analysis of product market structure based on involvement theory,Bayesian theory,consumer memory effect,sliding window effect and consideration set theory.This paper follows the research steps of theoretical research,model construction,data collection and data analysis.First of all,compared with page level analysis,this study collected product-level shopping data based on consumers' detailed decision-making sequence,and explores the relationship between consumers' purchase possibility on a product and the whole shopping decision-making process by combining the intervention degree theory.On the basis of product-level data,this study explored product attributes and optimization tools in a deeper level to discover their impact on consumer purchase behavior.Secondly,CBBI(consumer behavior based intelligence)model is proposed,the historical experience of consumers as prior information is considered to improve Bayesian model,that is,improved BLR model.Furthermore,the concept of sliding time window was added to describe the dynamics of consumer memory effect,and the global competitive market structure of product level is dynamically identified.Furthermore,the local sub-market competition structure of products was described by clustering and visualization methods Structure.Lastly,by introducing the consideration set theory,combining the sliding time window effect and the improved BLR model,we constructed the consumer consideration set,and further used the co-occurrence relationship of products to characterize the local product competition market structure with the consideration set.The results show that:(1)The purchase possibility of a product is closely related to the whole shopping decision-making process.Not all refinement tools have a positive and significant impact on consumers' purchase behavior,especially the sorting tools.In addition,consumers are highly concerned about product attributes.(2)Compared with the static Bayesian model,the BLR model is improved by adding the memory effect of consumers.Through MAE and RMSE analysis,the performance of the improved BLR model is better than that of the static Bayesian model.(3)The sales volume of products is time-dependent due to the memory effect of consumers.Combined with the robustness analysis based on other product categories,it can be determined that 5 days is the best time window for predicting product sales volume considering consumer memory effect on e-commerce platform,and it has generality in all categories.(4)Click behavior and browsing time have a significant negative impact on product sales;adding to favorites or shopping cart means that the product is more likely to have greater competitiveness in the market.In addition,consumers' behavior of removing products from shopping cart has a significant positive effect on product sales.(5)Based on the heterogeneous feedback of consumers,the proposed CBBI model can better describe the market structure(global)and sub-market structure(local)of products.(6)There are some limitations on consideration set construction by shopping cart or awareness set.Based on time window effect and improved BLR model,an effective method to construct consideration set was proposed.(7)Within the awareness set,the product distribution shows different regularities within and without consideration set.(8)The local market competition based on consideration set is closely related.Although some products have few competitors,they still own strong competitive advantages,and these products have a strong competitive relationship with a few or even one single product,occupying the main market.Most of the considered products are concentrated,which means that the competitors of these products are large,but occupy a temporary dominant position.It is found that with the sliding time window moving,the results have dynamic adaptability.