In the ever-evolving landscape of telecommunications, artificial intelligence (AI) stands as a disruptive and transformative force, reshaping how businesses interact with customers and streamline their operations. Embracing this technological revolution, companies like T-Mobile have pioneered innovative AI-driven solutions that elevate the user support ecosystem to new heights, delivering unparalleled customer experiences. By integrating AI to augment rather than replace human interaction, T-Mobile has followed a different path for customer care, blending advanced technologies with a human touch instead of replacing it. This approach not only enhances customer satisfaction but also drives operational efficiency, setting the stage for the future of telecom customer service.
Tanmaya Gaur, an expert in the field, shared with us his experiences in transforming user support applications across industries and revolutionizing customer interactions. In the telecom sector, recent technology advancements have resulted in significant customer experience innovations and optimization, specifically in utilizing data maturity and advances in AI/ML. At T-Mobile, Tanmaya’s focus has been on creating care experiences that surprise and delight customers. This includes using recommendation systems to better assist customer needs when support is requested as well as working towards capabilities for proactive resolutions to reduce the need for customer care contacts.
As the principal architect for customer care at T-Mobile, Tanmaya has added numerous such initiatives. He led the solution that set up the first Next Best Action pilot in customer care, a form of predictive care. By analyzing customers’ historical network, billing, and profile data, AI/ML suggests the next best action, such as a product or service the customer is not using or a new discount on a phone they’ve been considering. This ensures customers fully utilize their T-Mobile accounts, adding value in ways they might not realize. Tanmaya told us “There are plans to extend this capability beyond customer care into retail and into customer self-service interfaces”
Another major transformation he has been involved in is the development and integration of Expert Assistance tools for optimizing agent experiences. One example of this technology space is tools that process agent and customer conversations in real time, providing agents with relevant data points and account insights to address customer concerns effectively. This technology eliminates the need for customers to be put on hold while agents search for information, allowing agents to focus more on the human element of customer care. By partnering with AI Champions like Google, T-Mobile is working on augmenting these capabilities further.
A frequently overlooked challenge in these implementations is the actual presentation of dynamic AI/ML-driven insights to consumers of the data, be it customer service agents or end customers. There are challenges both in the integration of these dynamic insights into the care experience as well as getting enough adoption to be successful.
Tanmaya Gaur takes particular pride in the subtle, yet impactful innovation of the micro-frontend style Customer relationship management stack developed for the care agents at T-Mobile. Previous to the new solution, agents had to switch between multiple disjointed tools. Tanmaya and his team architected the new solution using micro-frontend technology to offer a more contextual, performant, and unified agent experience while remaining modular, maintainable, and agile for development and operational teams. The new cohesive micro-frontend-driven experience strategy is crucial in enhancing the efficiency and effectiveness of customer support operations.
The modular development paradigm championed by this tool allowed T-Mobile to seamlessly and quickly integrate and prove various AI/ML capabilities. These pilot experiences could be control tested with a small population of agents to prove value. The ones which were successfully were then be launched across all agents to deliver a more contextual, performant, and unified experience. This quick iterability between ideation, validation, and feedback played a big part in success of initiatives like the NBA. The composable underlying architecture is what allows for flexibility and helps deliver the overall goal of making the agent experience more dynamic and relevant to both the calling subscribers and the agents servicing those calls.
AI has significantly transformed user support applications across the globe, revolutionizing how businesses interact with customers. In the telecom sector, apart from the agent experience, these advancements have also focused on enhancing automation, and operational efficiency. At T-Mobile, the focus continued to be on creating the best customer service experiences while retaining the human element of the interaction. The other area of exploration with AI has been providing proactive resolutions to reduce the need for customer care contacts.
Tanmaya mentioned that he participated in several memorable initiatives within and outside his organization in this space, significantly impacting the realm of customer relationship management (CRM) and customer service management (CSM). A major enterprise initiative was the Next Best Action project, where they leveraged Pega’s Customer Decision Hub (CDH) to create a sophisticated recommendation system. CDH integrated hundreds of enterprise data signals from various systems within T-Mobile, ranging from simple data points like the customer’s current plan to complex metrics such as network health scores. Using AI and machine learning, the solution analyzed these signals to determine the next best action tailored to each customer’s unique needs and context. These insights were then presented to customer care agents with associated quick actions, requiring significant CRM integration and enhancements.
Another notable project that he was part of was the implementation of auto-summarization for customer care calls, significantly reducing the manual effort required by agents as well as optimizing call handling times. This initiative was part of a broader project known as Expert Assist. The work in making dynamic and actionable AI/ML data available within a CRM also led to the filing of several patents, underscoring the innovative impact of these projects.
These implementations have delivered significant business value at T-Mobile. The next Best Action (NBA) initiative at T-Mobile has delivered remarkable outcomes in measurable terms. Further mentioning the project, he shared a few insights, “These initiatives have been a key component of the 8-point increase in the Net Promoter Score and have boosted customer retention. T-Mobile achieved a postpaid churn rate of 0.96% in Q4 2023 and 0.87% for the entire year of 2023, marking the lowest churn rate in our history.” These achievements, along with other innovations, have contributed to T-Mobile winning 13 consecutive JD Power awards for excellence in customer care.
Tanmaya discussed the significant challenges he faced as the architect for the first implementation of the NBA pilot at T-Mobile. The primary objective was to collaborate with PEGA, a leader in AI/ML solutions for the telecom industry, to develop a system that could generate customer insights for upsell and retention purposes. This involved running ML models on customer data attributes to determine propensity scores for driving upsell and churn reduction experiences. Several key challenges arose during this process:
One major challenge was the dispersed nature of T-Mobile’s data across various enterprise systems. Predicting the data needs for the solution and ensuring that all necessary data was sanitized and available in near real-time for model training and execution proved to be critical hurdles.
Another significant challenge was ensuring compliance with multiple data privacy laws, such as CPNI and PCI. Adhering to these regulations required considerable development work, especially to comply with laws like the California Consumer Privacy Act (CCPA), ensuring that all data made available was in full compliance.
Additionally, establishing control groups to validate the solution’s results was a complex task. Creating control and treatment groups that were similar in all aspects to minimize the impact of external confounding variables required more effort than initially anticipated.
Addressing these challenges involved substantial work to ensure that all required data was consolidated into a common data lake. This foundational effort once completed provided significant flexibility and agility for the data science and customer care teams to develop new and more effective models. The overall solution led to increased agent satisfaction, demonstrating the success of this transformative initiative.
He talked about the several innovations in delivering these solutions which have been submitted as research papers. He also talked about the two patents specific to this domain, which are in review by USPTO. The first patent involves a rewrite of the CRM and CSM into a modular, composable UI that allows T-Mobile to provide dynamic experiences to their agents. The second patent focuses on presenting dynamic UI experiences based on AI/ML insights in the CSM troubleshooting space.
From the standpoint of an experienced professional in this field, Tanmaya Gaur observed that artificial intelligence and generative AI are among the most influential emerging trends in customer support. Many businesses are now deploying software bots to serve customers. However, T-Mobile has adopted a distinct approach by using AI to enhance the customer experience rather than intercepting the caller before reaching a human representative. Looking ahead, T-Mobile aims to offer more personalized, data-driven experiences to both customers and support representatives. They plan to utilize generative AI to transition towards more proactive rather than reactive customer care. After all, no customer truly wants to have to call customer service.
One crucial but often overlooked aspect he talked about is the “last mile” challenge. Many AI/ML enterprise solutions fail due to reasons beyond the efficacy of the solution. Critical aspects that drive failure are lack the proper buy-in from stakeholders, not addressing the right business challenges, training on inaccurate or unreliable data, or because they are not well understood by the end users. Ensuring that these models and solutions are accurately targeted, well-supported, and effectively integrated into the user experience is critical for achieving successful results.
The work with the NBA and ATLAS has garnered attention in various media outlets. Notable publications have highlighted the advancements, including articles such as “T-Mobile Will Use Google Cloud for Customer Care,” “T-Mobile Customer Decision Hub,” “At T-Mobile, Machine Learning and AI Are Transforming Customer Service,” and “Tex-Tech: How Technology Helps T-Mobile Perfect Customer Care.” These pieces showcase the significant strides made in integrating cutting-edge technology to enhance customer service and operational efficiency.