Method and system for identifiying subscribers likely to cancel subscriptions

Abstract

The invention is a method for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions, performed by a system comprising a subscriber tracking server (100) and subscriber-specific user terminals (200) that enable subscribers to access conntent of the content provider. It is characterized by collecting subscriber data, including information about subscribers and recorded actions taken by subscribers during subscription, by the subscriber tracking server (100), pre-processing subscriber data by the subscriber tracking server (100), achieving a first artificial intelligence model trained with data on subscribers who have canceled their subscriptions and those who are still subscribed, and, upon receiving subscriber data as input, that generates a loyalty prediction expressing the probability of the subscriber with the subscriber data it receives as input canceling the subscription, applying the received subscriber data to the artificial intelligence model to receive the loyalty prediction as output, sending the loyalty prediction content to a content provider sever.

Description

[0001] DESCRIPTION METHOD AND SYSTEM FOR IDENTIFYING SUBSCRIBERS LIKELY TO CANCEL SUBSCRIPTIONS

[0002] TECHNICAL FIELD

[0003] The invention relates to a method for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions, performed by a system comprising a subscriber tracking server and subscriber-specific user terminals that enable subscribers to access content of the content provider.

[0004] PRIOR ART

[0005] Pay-TV and similar content providers experience customer churn during their subscription services. This churn results from subscribers discontinuing their subscriptions due to dissatisfaction with the service for various reasons. Current methods generally rely on manual and statistical analyses to conduct customer churn prevention efforts. These analyses are based on customer complaints and feedback, and are usually communicated by customers via e-mail or through forms on the websites. In this process, receiving and evaluating complaints requires intensive manual processes, leading to a time-consuming and inefficient process. Therefore, an automated, more efficient, and faster solution is necessary in this field.

[0006] As a result, all the above-mentioned problems have made it imperative to make an innovation in the relevant technical field.

[0007] SUMMARY OF THE INVENTION

[0008] The present invention relates to a method and system for eliminating the above- mentioned disadvantages and bringing new advantages to the relevant technical field.

[0009] An object of the invention is to introduce a computer-based method for automatically detecting subscribers who have similar patterns of behavior to subscribers who have canceled subscriptions. Another object of the invention is to introduce a method for subscribers to communicate their complaints to the content provider in a facilitated manner.

[0010] In order to achieve all the purposes mentioned above and that will emerge from the detailed description below, the present invention is a method for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions, performed by a system comprising a subscriber tracking server and subscriber-specific user terminals that enable subscribers to access content of the content provider. Accordingly, it comprises the steps of collecting subscriber data, including information about subscribers and recorded actions taken by subscribers during subscription, by the subscriber tracking server, pre-processing subscriber data by the subscriber tracking server, achieving a first artificial intelligence model trained with data on subscribers who have canceled their subscriptions and those who are still subscribed, and, upon receiving subscriber data as input, that generates a loyalty prediction expressing the probability of the subscriber with the subscriber data it receives as input canceling the subscription, applying the received subscriber data to the artificial intelligence model to receive the loyalty prediction as output, sending the loyalty prediction content to a content provider server. Thus, it is ensured that subscribers who have similar behavior patterns with the former subscribers who have canceled subscription are identified and measures are taken to prevent their departure.

[0011] A possible embodiment of the invention is characterized by allowing the user to fill out a complaint form by the user terminal and sending the filled complaint form to the subscriber tracking server, determining whether the complaint form has an anomaly using an anomaly detection method by the subscriber tracking server, and ignoring the complaint form if it is determined that the complaint form has an anomaly. Thus, it is ensured that subscribers can easily report complaints and the attempts of subscribers who try to benefit from the system in bad faith are prevented.

[0012] Another possible embodiment of the invention is characterized in that if it is determined that the complaint form does not have an anomaly, it comprises the steps of classifying the complaint through a natural language processing model, and adding the classified complaint to the subscriber data.

[0013] BRIEF DESCRIPTION OF THE DRAWING Fig. 1 shows a representative view of the system implementing the method.

[0014] DETAILED DESCRIPTION OF THE INVENTION

[0015] In this detailed description, the subject of the invention is explained by way of example only for a better understanding of the subject, which shall not create any limiting effect.

[0016] The invention is a method for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions, performed by a system comprising a subscriber tracking server (100) and subscriber-specific user terminals (200) that enable subscribers to access content of the content provider.

[0017] Referring to Fig. 1 , the system consists of a user terminal (200) that serves content by a content provider system. The content provider system (500) provides content to the user terminal (200) in a streaming, on-demand, or broadcast format via a communication network, such as the Internet. Said content can be video content. This content service can be, for example, a service known as Pay-TV in the technique.

[0018] The user terminal (200) comprises a communication unit (230). Through the communication unit (230), the content provider can receive the content from the system. A communication unit (230) can be a device that can be connected to a wireless or wired network or to a cellular network. The user terminal (200) comprises a processor unit. The processor unit (210) performs the method steps of the invention and playback of content by executing software containing functional modules consisting of command lines recorded in a memory unit (240). The user terminal (200) also comprises a user interface (220). The user interface (220) may comprise audio and video output hardware controlled by the processor unit to enable content playback, and input hardware to enable user input. Without limitation, the user terminal (200) can be a smartphone, a tablet computer or any general purpose computer, etc.

[0019] By tracking the recorded actions of subscribers, the subscriber tracking server (300) identifies members who are likely to cancel subscriptions thanks to artificial intelligence models, and also ensures that member complaints are evaluated in an accelerated manner and taken into account in the evaluation of subscriber satisfaction. In more detail, the subscriber tracking server (100) comprises a server processor unit (110). The server processor unit (110) executes a subscriber tracking module (120) saved in a server memory unit (130) and performs the method steps of the invention. The subscriber tracking module (120) comprises a data collection module (121 ). The data collection module (121) receives the recorded actions of registered subscribers. The subscriber tracking module (120) obtains data using, for example, personal information entered by the subscriber during registration, content consumption actions performed by the subscriber during subscription, the registered subscription fee payment pattern, relevant data from geospatial data sources (400), logs from the user terminal (200) of another type, such as a set-top device that the user is using at the same time and to which content is delivered by a content provider system for example via terrestrial or satellite or cable broadcasting, logs from the user terminal (200), and stores this data as subscriber data.

[0020] A data processor module (122) subjects the subscriber data to a pre-processing process. The data processor module (122) makes the collected data suitable for training, deploying data-driven machine / deep learning models or for generating inputs for these models. The data processing module (122) is prepared. This involves removing missing or erroneous data from the data and converting the data into a format that can be used by machine learning models. Processes such as attribute extraction can be applied. This type of processes is not described in more detail here as it is well known in the art.

[0021] An artificial intelligence module (123) comprises a first artificial intelligence model trained with data on subscribers who have canceled their subscriptions and those who are still subscribed, and, upon receiving subscriber data as input, that generates a loyalty prediction expressing the probability of the subscriber with the subscriber data it receives as input canceling the subscription. When a subscriber's subscriber data is given as input to this artificial intelligence module (123), it generates a loyalty prediction as an output. Said loyalty prediction here can be graded according to whether the user belongs to the class of subscribers who show tendency of terminating the subscription or to the class of subscribers who do not show tendency of terminating the subscription. The artificial intelligence model can be, for example, a deep learning model. The artificial intelligence model is implemented in more detail using the deep graph-based similarity model.

[0022] In a possible embodiment of the invention, the user tracking module also comprises a communication module. Said communication module sends messages to subscribers classified in the category of users who show tendency of terminating the subscription using instant communication methods such as SMS, push notifications, and in-app messages. The communication module (124) also performs the functions of receiving customer complaints and sending notifications to the customer regarding this.

[0023] In a possible embodiment of the invention, the user terminal (200) is configured to allow the subscriber to fill out a complaint form. For example, it is possible to enter a complaint letter through an application or to create a complaint form by selecting the appropriate ones from the predetermined options. The created complaint form is received by the subscriber tracking server (100). In this embodiment, the data processor module (122) determines whether the complaint form has an anomaly using an anomaly detection method by the subscriber tracking server (100). If it has an anomaly, this form is ignored. The anomaly here may be, for example, forms that do not contain meaningful data created by the subscriber for trial purposes, many forms sent with the same content one after the other, etc. The data processor module (122) classifies the complaint by means of a natural language processing model if it determines that the complaint form does not have an anomaly. The data processor module (122) adds this classification to the subscriber data and sends it to the artificial intelligence module (123).

[0024] A graph database (125) is provided within the subscriber tracking module (120). The graph database (125) stores information about the relationships between subscribers and the pay-TV company, as well as the relationships between the customers themselves. This information can be collected from the pay-TV provider's customer records and social media data. The collected information can be added to the subscriber data and given as input to the artificial intelligence model, and the artificial intelligence model can also be trained with this type of data.

[0025] The method of the invention is performed by the system detailed above The subscriber tracking module (120) may comprise a quantum computing module (not shown in the figure). The quantum computing module functions in place of the artificial intelligence module (123) to train and deploy data-driven machine / deep learning models to predict customer churn and simulate different customer churn scenarios. This module performs processes using quantum computing methods known in the art. The server is also suitable for quantum computing. A combination of data-driven machine / deep learning algorithms, geospatial data, graph-based methods, and quantum computing methods can be used to improve the accuracy of subscriber churn prediction. For example, the system is able to use data-driven machine / deep learning algorithms to identify the most important features for customer churn prediction, and then use graph-based methods to model those features. The system can then use quantum computing methods to train a more accurate model using these features and relationships.

[0026] The communication module can use artificial intelligence methods to establish personalized communication with customers to prevent customer churn. For example, the system may send special offers or promotions based on customers' interests and past purchases. This helps customers stay loyal to pay-TV services.

[0027] The method performed by the system, the details of which are described above, comprises the following steps in a main embodiment of the invention:

[0028] - collecting subscriber data, including information about subscribers and recorded actions taken by subscribers during subscription, by the subscriber tracking server (100),

[0029] - pre-processing subscriber data by the subscriber tracking server (100),

[0030] - achieving a first artificial intelligence model trained with data on subscribers who have canceled their subscriptions and those who are still subscribed, and, upon receiving subscriber data as input, that generates a loyalty prediction expressing the probability of the subscriber with the subscriber data it receives as input canceling the subscription,

[0031] - applying the received subscriber data to the artificial intelligence model to receive the loyalty prediction as output,

[0032] - sending the loyalty prediction content to a content provider server (300). The content provider server can authorize transactions such as establishing special communication paths for relevant subscribers, sending offers, etc.

[0033] In an alternative embodiment of the invention, the method includes the following steps in addition to the following main embodiment:

[0034] - allowing the user to fill out a complaint form by the user terminal (200), and

[0035] - sending the filled complaint form to the subscriber tracking server (100),

[0036] - determining whether the complaint form has an anomaly using an anomaly detection method by the subscriber tracking server (100),

[0037] - ignoring the complaint form if it is determined that the complaint form has an anomaly.

[0038] Alternative embodiment may also include the following steps:

[0039] - if it is determined that the complaint form does not have an anomaly, classifying the complaint by means of a natural language processing model,

[0040] - adding the classified complaint to the subscriber data.

[0041] The scope of protection of the invention is specified in the appended claims and cannot be limited to what is described for illustrative purposes in this detailed description. It is clear that a person skilled in the art can produce similar embodiments in the light of what is explained above, without deviating from the main theme of the invention.

[0042] REFERENCE NUMERALS GIVEN IN THE DRAWING

[0043] 100 Subscriber tracking server

[0044] 110 Server processor unit

[0045] 120 Subscriber tracking module

[0046] 121 Data collection module

[0047] 122 Data processing module

[0048] 123 Artificial intelligence module

[0049] 124 Communication module

[0050] 125 Graph database

[0051] 130 Server memory unit

[0052] 200 User terminal

[0053] 210 Processor unit 220 User interface

[0054] 230 Communication unit

[0055] 240 Memory unit

[0056] 300 Content provider server 400 Data source

[0057] 500 Content provider system

Claims

CLAIMS1 . A method for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions, performed by a system comprising a subscriber tracking server (100) and subscriber-specific user terminals (200) that enable subscribers to access content of the content provider, characterized in that it comprises the steps of;- collecting subscriber data, including information about subscribers and recorded actions taken by subscribers during subscription, by the subscriber tracking server,- pre-processing subscriber data by the subscriber tracking server,- achieving a first artificial intelligence model trained with data on subscribers who have canceled their subscriptions and those who are still subscribed, and, upon receiving subscriber data as input, that generates a loyalty prediction expressing the probability of the subscriber with the subscriber data it receives as input canceling the subscription,- applying the received subscriber data to the artificial intelligence model to receive the loyalty prediction as output,- sending the loyalty prediction content to a content provider server.

2. A method according to claim 1 , characterized in that it comprises the steps of;- allowing the user to fill out a complaint form by the user terminal, and- sending the filled complaint form to the subscriber tracking server (100),- determining whether the complaint form has an anomaly using an anomaly detection method by the subscriber tracking server,- ignoring the complaint form if it is determined that the complaint form has an anomaly.

3. A method according to claim 2, characterized in that it comprises the steps of;- if it is determined that the complaint form does not have an anomaly, classifying the complaint by means of a natural language processing model,- adding the classified complaint to the subscriber data.

4. A system comprising a user terminal (200) and a subscriber tracking server configured to perform the method according to any one of the claims between claim1 and claim 3 for classifying subscribers of a content provider based on their likelihood of canceling their subscriptions.

Images