There is major buzz around artificial intelligence (AI) across the communications industry, but it is often discussed in conceptual or deeply technical terms. Real examples are relatively rare; even rarer is the application of multiple examples of AI and associated technologies, such as machine learning and big data analytics, to address issues impeding revenue for communications service providers (CSPs). TM Forum’s Empowering business assurance with AI Catalyst does exactly that.
Innovative digital services, such as smart home, e-health, connected car, etc., bring with them potential new revenue streams. However, there is often a disconnect between the high-level goals and activities of digital transformation and the specific IT issues experienced by CSPs when they attempt to create new solutions. British Telecom, Deutsche Telekom, Orange and Telia are championing the Catalyst project to address this disconnect. Participants in the project include Amdocs, Brytlyt, Subex and WeDo Technologies.
Watch Amdocs’ Dr. Gadi Solotorevsky explain the project:
What are the challenges?
The CSP sponsors identified 17 specific issues that are causing difficulties in their businesses as they seek to launch new services, increase operational flexibility and improve customer experience. In the first phase of the project, the participants chose to address seven of those challenges with the rest to follow in later phases. In all cases the team is using big data analytics and AI, with different algorithms and models implemented to deliver the functionality required. In addition, behavioral analysis, natural language processing and sentiment analysis were used for specific uses cases outlined below:
- Identity authentication – this use case was developed in response to the increasing requirement for customers to activate new services and manage their accounts in real time via online portals and apps. Such methods are fast replacing the traditional method of calling a contact center to make changes to an account. Potential for identity fraud will increase if identities cannot be authenticated quickly. The Catalyst team used AI and big data analytics to search and cross match identity data from internal and external sources in order to ensure that the customer is who they should be.
- Customer credit – there is a requirement to shift from a one-time check at the point of signing up to ensuring that the customer has the right credit profile to sign up for a range of new services and functionality whenever they want to. This is essential for CSPs looking to diversify and expand their service portfolio, build platform businesses and provide greater customer centricity as part of the digital transformation. The Catalyst solution included big data analytics and artificial intelligence, using web and social media sources as well as internal data to enhance credit checking and updating.
- Internal fraud prevention – CSPs have developed and continue to improve anti-fraud processes designed to combat internal fraud. However, as the number of operations increases and services multiply, there are more possibilities for fraud from inside the CSP. The Catalyst team used AI and machine learning to automate functions and eliminate possibilities for fraud and used big data analytics to pro-actively identify incidents of fraud.
- Preventing provisioning failures – failures and delays in the provisioning process, particularly those associated with incorrect or incomplete technical information for installation, will have a significant impact on the customer’s experience and their view of the service. The team applied machine learning and AI to anticipate and prevent problems at the moment the order is placed, permitting the CSP to take proactive steps to improve provisioning time and reduce failures caused by systems, processes and human error.
- Detecting impersonation – working hand-in-hand with identity authentication, the countering of deliberate, large-scale attempts to impersonate users becomes more important as the number and value of CSPs’ services increase. Point identification techniques are limited and impersonation detection is often too reactive, identifying issues too late. The team used AI and behavioral analysis to identify characteristics of scale impersonation and possible fraud with greater accuracy and immediacy.
- Customer experience improvement – the customer’s interaction with a CSP is shifting from a single point of contact in a call center, to multi- and omnichannel options. However, information about the customer gathered from these various channels are not linked together. The Catalyst uses big data analytics, natural language processing and sentiment analysis, to bring together information held in separate systems and social networks and provide a better customer experience. For example, the team is not only able to combine information from the various channels about an experience but also measure the level of pleasure or displeasure using sentiment analysis.
- Partner fraud prevention – fraud prevention is primarily inwardly focused, with interconnection and roaming the only real examples of services that involve partners. However, with expansion of the digital ecosystem, CSPs must understand and identify fraudulent activities beyond their own networks. During the Catalyst demonstration, AI and machine learning were used to quickly identify and stop attempts to bypass traditional fraud-detection algorithms initiated by or within partner domains. Machine learning was also applied to automate interactions between CSP and partner domains, such as additional service activation, and therefore limit possibilities of fraud instigated by employees of partners.
For each of the use cases, multiple complementary technologies (for example, big data analytics, AI and machine learning) were employed in combination. This was not a case of evaluating a technology to see how it could be used, but rather using the technologies that vendors have developed as a toolkit. When applying these technologies, the Catalyst team also used TM Forum’s Analytics Big Data Repository (ABDR), applying some of its assets to develop the architecture.
Seeing is believing
The Catalyst was demonstrated at TM Forum Live Asia 2017 in Singapore and took attendees through a complete customer journey. The customer, Paul, subscribes to a smart home service. The demo operates in a live commercial system and follows Paul through his journey to demonstrate the various business assurance use cases in action. The graphic below shows the role each participant played.
Attendees were shown how AI, machine learning and big data analytics technologies can be used to empower the business assurance process, which includes improving customer experience and revenue assurance. Watch the team’s video from Singapore below.
Initial outcomes and next steps
A main goal of the Catalyst is to validate the potential for machine learning and AI in business assurance. It is all about concrete steps, not theoretical slideware, and the learnings from this project are already being fed back into the relevant TM Forum collaboration teams. Specifically, eight new entities were defined and contributed back to the TM Forum Analytics Big Data Repository (ABDR):
The success of the first phase of the project means that the team is quickly moving on to address more of the 17 issues originally identified by the Catalyst’s CSP sponsors, such as how to use specific customer-service data and how to welcome customers home and automatically start appropriate services. The next phase will be demonstrated at TM Forum Digital Transformation World in Nice in May, 2018. This phase will also enhance the architecture, add new use cases, and provide more contributions back to TM Forum’s assets.
If you’d like to get involved in this Catalyst project, please contact Tania Fernandes via [email protected].