|← Momentous Developments||Air Canada →|
According to Griffin (2005), CRM primarily addresses the customer relation issue. However, customer selection is but just an umbrella covering many smaller related concerns. In the current first moving business world, just carrying out business with a big number of customers is a sufficient way of establishing and sustaining a competitive edge. A widespread thumb’s business rule observes that about 20% of customers a firm generate about 80% of its profit. Companies like Amazon.com have time and again used CRM tools for separating good customers from good ones. Good customers at Amazon.com are those ones boosting profits by buying large quantities an expensive merchandize the company offers. The “bad” ones on the other hand exploit programs for price matching, take advantage of policies of return and abuse rebates and in the end, this behavior hurts the company’s profitability. The tool helps to eliminate most of the mystery surrounding consumer behavior. This tool can be used to make multi-dimensional databases for customers, loyalty programs, relationship programs and recommendation systems.
Swift (2000) affirms that there are several data mining technologies and tools that a strong CRM package uses. For example, associations rules can be used to serve profitable clients through analyzing past buy behavior to foretell future purchases. Consequently, stores can alter their product showcases to collocate profitable consumers’ desired items. Amazon.com also uses association rules to highlight the services and products that may gain from incentives.
Another CRM tool used by Amazon.com is clustering. Through this method, a firm’s big customer base is segmented into many small groups with identical prospects. The CRM user is therefore allowed to create more efficient marketing campaigns which are particularly designed for targeting important customers and give them a service or product that they will probably purchase (Griffin, 2005).
Other aspects that CRM incorporates are various models of classification entailing collaborative filtering, neural networks and decision trees. All the above mentioned techniques and technologies enable companies like Amazon.com to separate profitable customers from bad ones and to provide for them accordingly.
Uses of CRM at Amazomn.com
There are several ways in which Amazon.com and other companies use CRM. One of such uses is regression and scoring models to establish the phone that belong to businesses within their database. In the end, the outcome will be a technique that can be used for targeting profitable home business fragment, one that ca not be easily identified and served by traditional methods (Swift, 2000).
CRM has widely been used for acquiring, sustaining and understanding the customers who are most profitable. Amazon has been allowed growth because of CRM’s ability to simultaneously offer its customers treatment and experience that are generally linked with smaller businesses
Griffin (2005) observes that using CRM has some problems associated with it. For instance, the malicious cycle and the vicious cycle of CRM arise. Malicious cycle of CRM happens when the use of information and acquired with methods of data mining influences and manipulates consumer behavior. This will allow a customer’s weakness to be exploited if adequate data concerning them is obtained. On the other hand, the vicious cycle of CRM is associated with customer predictions that are erroneous (Swift, 2000). At some point, even the most precise CRM system may and will generate false predictions. A system may assume to know a customer when it actually doesn’t. As a result, valuable customers may be recommended offensive materials. This problem is almost impossible to be fixed especially if the systems being used rely on inherent recommendation systems and techniques of collaborative filtering.
Amazon.com employs item-to-item collaborative filtering for generation of its customers’ personalized recommendations. This method begins with data concerning history of each customer and their personal preferences. From here the method engages a distance function-like the one used for clustering method, to find and group customers enjoying the same kind of products (Griffin, 2005). Consequently, there is the use of votes to append weights to distances, thus products that are favored by the most identical customers are more dominant to recommendations of Amazon.com. The use of collaborative filtering crafts an exceptional implicit system of recommendation: there is no need of customers deliberately disclosing their preferences so that they can get accurate recommendations but Amazon.com will use their historical behavior to automatically spawn precise recommendations.