Claim Prediction in TELCOs

Machine Learning and DataRobotics to generate value in the Telco Industry: churn risk management tools.

Match-Up

As-is Scenario

Today, Telco Service Providers compete in terms of technological and business capacity, in order to guarantee customers an excellent Quality of Experience. In fact, a satisfying use of data and voice services allows each of us to improve our everyday life, making it possible to enjoy indispensable digital services to the full. Mobile payments, smart home or remote working services, gaming or virtual and augmented reality would be a futuristic idea if the performance of telco providers were not fully reliable, or if the performance of network assets distributed throughout the territory were not adequate to meet users’ traffic requirements.

In this context, the adoption of Machine Learning & Process Automation solutions enable countless benefits that go far beyond the mere efficiency of the Operations, also affecting the effectiveness of Customer Retention activities and the innovation of the whole business model, in order to develop new customer-centric services capable of impressing customers with increasingly personalised value propositions.

To-be Scenario

In the TO-BE scenario, Telco Service Providers have a system at their disposal that adapts to different application contexts and which:

  • Is connected to a data layer specialised in Telco data, with the main functionality being the integration and pre-processing of Network Performance, Asset Inventory and CRM data, retrieved from different databases.
  • Receives accurately pre-processed input data in order to immediately provide significant insights on each customer at risk of churning.
  • Offers a “predictive analytics” feature provided by a Machine Learning model designed to return predictive classification output labels, in other words capable of assigning a “churn score” to each customer analysed (“Claim Prediction Tool”).
  • Provides “prescriptive analytics” features, carried out by a Recommendation Engine which generates suggestions on actions to be taken, or assigns a “best action” to the type of criticality identified by the customer being analysed (with specialised future application for the management/display of recommended actions) (“Best-action Tool”).
  • Integrates the Machine Learning output (“churn score” and “best-action”) to the other incoming customer profiling data.
  • Presents an interactive front-end, in other words, a dashboard application that operates as a unified view on all the information managed (from the data stored in upstream business databases, to data generated through automatic learning).
  • Makes it possible, as a whole, to manage customers at risk of churning, by proactively taking action to anticipate possible customer issues such as the opening of tickets, etc.
  • Guarantees an analysis of the main characteristics and needs of the customer base through the use of specific views by customer segment.
  • Generates reports containing details and KPIs, including an assessment of the “Quality of Experience” with respect to deviations from predefined “Quality of Service” thresholds.
  • Is able to generate, on request, a master log file of the main network failures that have the most negative impact on the customer experience.
  • Makes it possible to evaluate the best actions recommended with regard to customer retention and incident resolution for each user case, both on the basis of historical post-sales actions undertaken up to that time, and on the basis of new measures specifically studied to ensure customer loyalty and prevent customer dissatisfaction with respect to a certain type and range of clientele.

FOcus on

In the Telco Industry, a “churn risk management” tool is essential for ensuring customer centricity and an adequate level of profitability. Proactively overseeing customers and their behaviour can help support the company in containing costs and preventing risks.

The “Claim Prediction Tool” is a tool based on a predictive Machine Learning model that automatically assigns a “Churn Score” to each user in the customer base, enriching their profile with information on the customer segment to which he/she belongs. Customer profiling ranges from KPIs for the provision of the service included in the commercial offer, to the supply of devices used by the customer, without ignoring the history of the relationships that have characterised the customer-service provider relationship up to that point in time (previous reports of dissatisfaction, changes to the rate plan, etc.).


Critical aspects
In the Telco Industry, the enabling sector of the Fourth Industrial Revolution, the more or less latent signals left by customers at risk of churning reside in the various data sources which companies have at their disposal, due to the very nature of the service offered to end users. Service delivery in the Telco sector is based on a technical layer – made up of network assets, territorial points-of-presence and transmission equipment, etc. – followed by a layer of configuration and coupling between the network and the service, through which the offer is provided. Re-constructing the network fruition chain and the touchpoints between the service provider and the user therefore means interfacing with several different IT systems and querying different Data Warehouse families to extract the relevant data.

Moreover, obtaining information from the different distributed data sources entails carrying out accurate calculations that bring to light the most significant insights hidden in the network usage and asset capacity patterns. Predicting the Telco customer's risk of churning in advance, based on its consumption behaviour and on the actual network performance observed by each user therefore represents a complex activity, which requires a substantial use of resources and corporate commitment in terms of customer centricity.

A “churn risk management” tool is essential for ensuring customer centricity and an adequate level of profitability. Proactively overseeing customers and their behaviour can help support the company in containing costs and preventing risks. A dissatisfied customer can indeed become the worst detractor of the company’s brand, amplifying the effects of their frustration, interrupting the process of word-of-mouth marketing to the so-called 3Fs (Friends, Family and Fools) and in the worst case scenario, sharing with FFFs all the details of a mediocre consumer experience. The same churn risk management activity enables Telco service providers to generate new revenue streams: cultivating an already consolidated relationship predisposes the customer to becoming the company’s best partner, creating an open and transparent relationship.

The customer who is “listened to” is more inclined to freely share tips and suggestions relating to commercial packages and CRMs, to accommodate cross-selling and up-selling initiatives, and to pay premium prices for offers, etc. Conversely, an ignored customer is less inclined to authorise the company to study their behaviour, to agree to the processing of personal data for commercial purposes, to co-design the “ideal service” with the company aimed at their individual satisfaction or for that of the relevant market cluster (peer-to-peer collaboration).

The “Claim Prediction Tool” as a solution
The added value of the ClaP system manifests itself in terms of:

  • The ease of use of the tool, to be integrated into the work tool-kit available to proactive caring process owners.
  • Immediate access to the relevant information, offering an integrated and unified view of data collected from different sources (databases and IT systems used by Telco Service Providers).
  • Information enrichment of every customer profile to support risk management activities;
  • Instant detection of risk of churning with adequate advance notice of the event with respect to its occurrence;
  • Real-time recommendations of “best actions” pertaining to issue resolution and customer retention;
  • Independence from platforms;
  • Flexibility of the usage model (cloud-based or residing on the company’s servers);

The developed use case
The developed use case is inherent to the “Claim Prediction Tool”, in other words an application that provides a predictive analytics featurecharacteristic of a Churn Risk Management system. The “Claim Prediction Tool” is a tool based on a predictive Machine Learning model that automatically assigns a “Churn Score” to each user in the customer base, enriching their profile with information on the customer segment to which he/she belongs. Customer profiling ranges from KPIs for the provision of the service included in the commercial offer, to the supply of devices used by the customer, without ignoring the history of the relationships that have characterised the customer-service provider relationship up to that point in time (previous reports of dissatisfaction, changes to the rate plan, etc.). The information enrichment of the customer profile may be extended to include data on network assets, at the basis of the technical layer of Telco service providers and of the navigation and voice services experienced by current users.

The tool is envisioned as a work-tool for customer operations agents. The tool can be used as part of the Proactive Caring activities to identify – using an integrated view – which customers belonging to the existing customer base are at risk of churning. It also supports the option to download the list of possible customers likely to be dissatisfied, to whom the Machine Learning model has assigned a “Churn Score”, with a priority score based on the confidence assigned by the model to the probability of an actual occurrence. The customer list can be examined at will according to customisable filters, according to the specific needs of the investigation (focus on the type of offer to which the customer subscribes, on the customer's geographic location, certification tools, etc.). The operator can view a detailed profiling for each customer at risk of churning, with items ranging from CRM information, KPIs relating to the services provided and elements relating to individual provisions and service collaterals (APP with Over-the-Top commercial offerings, etc.).

The solution architecture
The architecture consists of the following technical components:

  • Data import connectors (from a Data Integration Layer or specific databases);
  • Software or Machine Learning platforms trained on predictive tasks, through specially selected and periodically validated algorithms;
  • Recommendation engines trained on master files containing caring actions that have already been undertaken, or of preferable interventions in the resolution of similar cases;
  • A web server for the user interface;
  • A reports generator;
  • An input/output information database, with additional filters if necessary.


DISCOVER ALL THE REPLY DATA ROBOTICS ACCELERATORS


Data Robotics Accelerator

Automated Invoice

Automated Invoice is the solution, also available as a service, which facilitates the automated management of the accounts payable process, from the posting phase to the reconciliation between invoices and purchase orders/delivery notes/receipts, highlighting the differences identified.

Data Robotics Accelerator

Brick Machine Learning

Brick Machine Learning is the solution that makes it possible to simulate various configurations used in automated production lines, in order to recommend the optimal mix of devices required to achieve the overall performance requested by the end customer.

Data Robotics Accelerator

Customer Recovery

Customer Recovery is the solution that faces the challenge of behavioral approach on credit risk management. The solution is developed on Microsoft Azure Machine Learning, the service that allows building and testing powerful cloud-based predictive analytics.

Data Robotics Accelerator

Employee Monthly Expenses

Employee Monthly Expenses is the solution that facilitates the automated creation of expense reports starting from the underlying cost items, quickly and without the need for manual intervention.

Data Robotics Accelerator

Know Your Orders

Know your Orders is the solution that makes it possible to create a simple interface which can be consulted by users using a natural language, thus facilitating access to information while ensuring consistency and accuracy.

Data Robotics Accelerator

Match-up

Match-up is an advanced tool for the analysis, reconciliation and matching of complex data (single and/or multiple). The use of this tool finds application in data-related processes.