Post by wekeve7933 on Dec 5, 2023 11:01:26 GMT 1
We have entered the era of technological over-promise, one where there would no longer be any limit to what innovation can bring. This outbidding is also a formidable marketing weapon: many companies today use the “ buzzwords ” of the moment and promise the moon overnight. And on the podium of these absolute promises, predictive marketing is in good place… A column by Jean-René Boidron , CEO of Kameleoon & former Vice-President of CroissancePlus I should feel ill-placed to offer any criticism of predictive marketing. My driving force is innovation; One of the specialties of my company – and the pride of the R&D teams – is predictive algorithms. They are in fact THE lever to increase the conversion of our customers.
Problem: this term is used indiscriminately. It is therefore necessary to clarify it for digital Special Data experts (e-retailers or media). The growing opportunities offered by technologies are leading to the craziest promises such as connecting Los Angeles to San Francisco by electromagnetic train in 30 minutes or colonizing Mars, all by 2025. This new era of absolute promises has also become a formidable marketing weapon: many companies no longer bother to explain what it is possible to do or not to do... Predictive marketing has been around forever In fact, we've been doing predictive marketing since marketing itself existed. We take into account past experiences to inform the future. Realizing that traffic to his ready-to-wear store is high on Saturdays and low on Mondays, the merchant predicts that it will be the same in the future and decides to recruit an additional salesperson on Saturdays and to stay closed on Mondays. Likewise, he predicts that some items will sell better in winter than in summer; the experience acquired will even allow him to refine his decision-making on a weekly basis to optimize his sales. As its customer base becomes larger, it will also begin to segment its customers to maximize its sales by offering each one a range that is most likely to meet their needs. What predictive analysis will bring…
Many manual segmentations are made on a single criterion: CSP, gender, loyal customer or not, etc. Others, more elaborate, will cross two criteria: CSP with gender, frequency of purchase with place of residence, etc. From 3 criteria, specialized software will be essential and beyond that it will be necessary to use specialized algorithms especially if it involves processing thousands of customers. With the massive increase in information and processing capacity, artificial intelligence will already make it possible to automate a set of tasks in an exhaustive and error-free manner. Better still, machine learning algorithms will constantly take into account changes in customer behavior and characteristics to redefine in real time who belongs to which segment and what their most likely future behavior is. They will identify trends, find correlations and “ make the data speak ” in order to maximize conversion. Ideally, this task could be carried out manually but it would require an army of analysts night and day to carry it out. Freed from the tedious processing of data, these machine-learning algorithms will allow the marketer to concentrate on their strategy and the actions they wish to carry out. …And what it won't lead to 1st reality: do not confuse automation and predictive It is important not to confuse predictive analysis with the “ simple ” automation that results from the implementation of “ manual ” rules. Defining a rule that pushes a discount coupon to a visitor who stays 3 minutes on a page is good but it's automation, not artificial intelligence. Dynamic traffic allocation in A/B testing is good but it’s also automation, not predictive.
Problem: this term is used indiscriminately. It is therefore necessary to clarify it for digital Special Data experts (e-retailers or media). The growing opportunities offered by technologies are leading to the craziest promises such as connecting Los Angeles to San Francisco by electromagnetic train in 30 minutes or colonizing Mars, all by 2025. This new era of absolute promises has also become a formidable marketing weapon: many companies no longer bother to explain what it is possible to do or not to do... Predictive marketing has been around forever In fact, we've been doing predictive marketing since marketing itself existed. We take into account past experiences to inform the future. Realizing that traffic to his ready-to-wear store is high on Saturdays and low on Mondays, the merchant predicts that it will be the same in the future and decides to recruit an additional salesperson on Saturdays and to stay closed on Mondays. Likewise, he predicts that some items will sell better in winter than in summer; the experience acquired will even allow him to refine his decision-making on a weekly basis to optimize his sales. As its customer base becomes larger, it will also begin to segment its customers to maximize its sales by offering each one a range that is most likely to meet their needs. What predictive analysis will bring…
Many manual segmentations are made on a single criterion: CSP, gender, loyal customer or not, etc. Others, more elaborate, will cross two criteria: CSP with gender, frequency of purchase with place of residence, etc. From 3 criteria, specialized software will be essential and beyond that it will be necessary to use specialized algorithms especially if it involves processing thousands of customers. With the massive increase in information and processing capacity, artificial intelligence will already make it possible to automate a set of tasks in an exhaustive and error-free manner. Better still, machine learning algorithms will constantly take into account changes in customer behavior and characteristics to redefine in real time who belongs to which segment and what their most likely future behavior is. They will identify trends, find correlations and “ make the data speak ” in order to maximize conversion. Ideally, this task could be carried out manually but it would require an army of analysts night and day to carry it out. Freed from the tedious processing of data, these machine-learning algorithms will allow the marketer to concentrate on their strategy and the actions they wish to carry out. …And what it won't lead to 1st reality: do not confuse automation and predictive It is important not to confuse predictive analysis with the “ simple ” automation that results from the implementation of “ manual ” rules. Defining a rule that pushes a discount coupon to a visitor who stays 3 minutes on a page is good but it's automation, not artificial intelligence. Dynamic traffic allocation in A/B testing is good but it’s also automation, not predictive.