Article proposed by Yassine El Idrissi, consultant in CRM.
One of the most relevant definitions of Customer Relationship Management (CRM) is “an approach that aims to identify, attract and retain the best customers in order to increase the company’s profitability or the value of its customer capital”.
In this article, we will discuss the use of Artificial Intelligence (AI) and Big Data to identify, attract, retain the best customers and increase profitability.
Identify the best customers
In the past, the identification of the best customers was done through the turnover that the brand or company achieves with them. The higher the turnover, the more important the customer is, which is true in absolute terms. In addition, in each client portfolio, there are profiles with high potential but which have been assigned to less profitable segments of the company, due to a segmentation based on a single variable, namely turnover. This creates an enormous opportunity cost for the company or brand.
This is precisely where Big Data and AI come in, to make it possible to identify these customers. Indeed, the multiplication of data sources allows this identification.
Let’s take the example of a bank that has an unsegmented population of customers, while their customer’s account shows that they occasionally make purchases in luxury stores or restaurants. There is a good chance that the actions taken by the bank with regard to this population will be far from retaining them if not the opposite.
Today, algorithms are able to detect this type of customer accurately and according to several variables.
Identification can also be done by analyzing customers’ shopping baskets in the store or on a website. For example, a supermarket can analyze its customers’ shopping cart to identify cross selling opportunities. Algorithms such as APRIORI or CARMA are able to give fairly precise association rules of type: the purchase of product A implies the purchase of product B, or the purchase of products A and B implies the purchase of products C and D… The application of these association rules make it possible to increase the value of the average basket and consequently increase the turnover.
Identification can also be done through social networks. Indeed, customers no longer hesitate to share their opinions about a brand, the analysis of these public notices in a rather precise way is now possible thanks to Natural Language Processing (NLP) and Text Analytics. These techniques make it possible to detect the reasons for loyalty and the causes of infidelity to a brand or company.
Attracting the best customers
Attracting customers usually involves marketing campaigns. In traditional marketing, the return on investment of these campaigns has been the subject of much debate, so optimization approaches have been developed through impact analysis.
To give an example, let’s go back to the supermarket that has developed cross selling opportunities through AI algorithms such as APRIORI or CARMA, thanks to these algorithms, the supermarket has discovered that the purchase of product A involves the purchase of product B, but there are customers who buy product A but do not automatically buy product B, and these are exactly the target customers of a marketing campaign for product B.
The story is not over here, because the feedback from the campaign will be used to learn new algorithms for even better targeting in the next campaign. In other words, once the campaign is over, the customers who responded positively will first provide a better understanding of why product A involves product B, then better target the customers of the next campaign, and finally increase the return on investment of marketing campaigns.
Retain the best customers
The cost of acquiring customers and increasingly important, so loyalty becomes vital for the company or brand, loyalty can be measured by, among other things, an attrition rate, also called the churn (rate of customers leaving the company or brand on behalf of the competition), the higher this rate is, the less we retain our customers.
Currently, using artificial intelligence algorithms, we are able to anticipate this attrition and determine its causes using decision trees or supervised learning models. For example, a supermarket determined a score combining three variables called RFM (R: recurrence of customer visit, F: frequency of purchases, M: amount of purchase) to anticipate customer attrition. Using supervised models, it was possible to determine at what threshold of the RFM score the client is likely to leave the company and what actions to take once this threshold is reached.
In addition to RFM scores, algorithms such as C5.0, CHAID allow to identify attrition rules according to several variables and thus anticipate churn and build customer loyalty.
To summarize, the contributions of big data and artificial intelligence to customer relations are no longer to be demonstrated, especially with the multiplication of data sources thanks to social networks. On the one hand, they make it possible to transform unstructured data (text, image, etc.) into structured data, and on the other hand, to develop models that make it possible to segment customers or predict their behaviour.