By Anton Gora, engineer and developer of new products at Cianet
Chinese strategist and philosopher Sun Tzu (544 BC to 496 BC) anticipated by more than two and a half hundred years the essence of this text. In an excerpt from his best-known work, The Art of War, he wrote: "Those who solve them before they appear" excel at solving difficulties. Absolutely distant from the possibilities that Artificial Intelligence (AI) and Machine Learning (ML) have brought us in recent years, he predicted the corporate trend towards reducing churning, loyalty and increasing the customer base of modern companies, with a difference: today you can do this in a fraction of the time that the Eastern thinker needed to achieve success - and without wasting resources.
There are already two distinct and very clear groups on the market: that of companies that are developing some form of technology based on ML / AI and that of those that use it to improve their core business. The input for this most of them already have: information. Now, it is necessary to evolve in the way of dealing with it. What does this data say? Who can sue them? More than providing the name, address and income of the contractor, they can show the reality of the business in an almost scientific way and allow the right decisions to be taken at critical moments.
Take integrators and internet providers (ISPs) as an example. The services offered by these companies are contracted through subscription. The customer base, usually in the thousands, prevents them from dedicating too much time to users individually, but any cancellation needs to be avoided (or reversed). How to identify which subscriber clusters have less fault tolerance? What actions can minimize the possibility of contractual terminations? Answering these questions is a way of predicting churn.
For this, the company must go through some steps: collecting data, uploading information to a forecasting service and developing trends. With a CSV file and online analytics services, you can get started. The churn models will be set up as decision trees, where each node will be associated with a question about a value that the customer perceives about the company. Each answer, which is positioned like the branches, leads to the “leaves” that are associated with the possibility of users leaving. From them, it is possible to identify the potential churn rate (cancellation rate). These numbers may represent more than the simple fact that certain customers have lost interest in being served by the company. They are an indication that the business has stopped growing. That is why it is so important to analyze them and, above all, to act.
The same innovation that allows it to be more analytical and assertive, favors the transformation of the relationship with the customer. With machine learning applied to the entire process, his journey within the company is more personalized, eliminating one of the most frequent complaints from service contractors: the low quality of service. Gradually, the machines will acquire the ability to understand the preferences of the public and, through interfaces such as chatbots and standalone applications, offer solutions to problems that have not yet happened. The surprise factor of fault prediction and contour will improve the experience, eliminating the need for interaction with poorly prepared and little engaged attendants.
All of this, however, does not completely eliminate the need for human beings in the process. Artificial intelligence allows machines and systems to learn, that is, we will still need someone to teach them. This loop will remain joining the two ends of the process. Only it will be much less embarrassed.
There are already two distinct and very clear groups on the market: that of companies that are developing some form of technology based on ML / AI and that of those that use it to improve their core business. The input for this most of them already have: information. Now, it is necessary to evolve in the way of dealing with it. What does this data say? Who can sue them? More than providing the name, address and income of the contractor, they can show the reality of the business in an almost scientific way and allow the right decisions to be taken at critical moments.
Take integrators and internet providers (ISPs) as an example. The services offered by these companies are contracted through subscription. The customer base, usually in the thousands, prevents them from dedicating too much time to users individually, but any cancellation needs to be avoided (or reversed). How to identify which subscriber clusters have less fault tolerance? What actions can minimize the possibility of contractual terminations? Answering these questions is a way of predicting churn.
For this, the company must go through some steps: collecting data, uploading information to a forecasting service and developing trends. With a CSV file and online analytics services, you can get started. The churn models will be set up as decision trees, where each node will be associated with a question about a value that the customer perceives about the company. Each answer, which is positioned like the branches, leads to the “leaves” that are associated with the possibility of users leaving. From them, it is possible to identify the potential churn rate (cancellation rate). These numbers may represent more than the simple fact that certain customers have lost interest in being served by the company. They are an indication that the business has stopped growing. That is why it is so important to analyze them and, above all, to act.
The same innovation that allows it to be more analytical and assertive, favors the transformation of the relationship with the customer. With machine learning applied to the entire process, his journey within the company is more personalized, eliminating one of the most frequent complaints from service contractors: the low quality of service. Gradually, the machines will acquire the ability to understand the preferences of the public and, through interfaces such as chatbots and standalone applications, offer solutions to problems that have not yet happened. The surprise factor of fault prediction and contour will improve the experience, eliminating the need for interaction with poorly prepared and little engaged attendants.
All of this, however, does not completely eliminate the need for human beings in the process. Artificial intelligence allows machines and systems to learn, that is, we will still need someone to teach them. This loop will remain joining the two ends of the process. Only it will be much less embarrassed.
EN 
PT 
