Pros and cons

Why do companies need to predict their consumers behavior?

Now more and more companies try to use mined and analyzed data to predict consumers behavior to a produce: the aim? Maximizing gains while minimizing spending which are partly caused by the ignorance of consumers’ needs.

Companies which are specialized in modeling consumer profiles (epsilon, Acxiom) help sellers to target their audience more efficiently. (daddy company example).

In other areas such as ecology, Data mining, deep learning and algorithms help companies to predict consumer behavior: In Germany, the German Research Centre for Artificial Intelligence started last year a project to develop an algorithm for forecasting the production and consumption of renewable energy based on the principle of machine learning.

Consumer behavior prediction can allow companies to grow and make their value on the market evolve. Netflix example: years of data collection managed Netflix to move from a simple streaming platform to one of the biggest content creators. Netflix succeeded in understanding and predicting its users behavior to TV shows or movies and create a capital gain and prevail itself on the market.

Prediction by deep profiling and excesses

To predict consumer behavior as Netflix does for these recommendations, a huge amount of data is needed. Data profiling is therefore necessary to design a sufficient data quality and to analyze them.

Different techniques of data profiling:

• Completeness Analysis: identify the values which have no business significance and mark them “default values”
• Uniqueness Analysis: find how many distinct values exist and check duplicate data
• Pattern Analysis: analyze the evolution of customer demand over a specific time horizon

Marketing collaboration with the customer can result in promotions for customers, knowledge of the ideal time of launching a product…

At the corporate level, predicting customer behavior will better plan and forecast supply and replenishment. For example, Amazon prepares order before the customer has passed it by modeling the data: past orders, search history, wish list…

However, there is a problem we cannot avoid: the intrusion into consumer privacy. The desire to collect more and more accurate data is pushing companies to interfere in their private lives. For example, the “pregnancy predictability score”: it is the experience of a father who did not know that his daughter was pregnant but who received discount coupons for baby products.

Another problem is that consumers give their data without knowing where it goes. It is the example of the Cambridge Analytica scandal.