AI Use Cases in Telecom: Churn, Network, Fraud and Customer Service

Data Science Innovators Conference 2023AI Use Cases in TelecomWatch the full session · 23 min

In a session at the Data Science Innovators Conference 2023, Dr. Frances Boykin, director of advanced analytics at AT&T, surveyed where telecom companies use AI and machine learning, from churn prediction and network planning to fraud detection and customer service. Boykin said she works with teams focused on AI, machine learning and data implementation, and that telecom has engineers, marketers, customer experience, sales and service teams that both produce and use data. Her talk was followed by a long chat Q&A.

Where can AI help telecom companies?

Boykin said AI helps telecom companies in three broad areas: automating tasks, analyzing data and improving customer service. She grouped the challenges into customer engagement, infrastructure and products, and operations. The topics ranged from product recommendations and data monetization to network and capacity planning, churn analysis, proactive customer care, fraud, ordering and logistics, and fault management with root cause analysis.

How is AI used for loyalty, network quality and failure prediction?

On customer loyalty, Boykin said AT&T has done churn analysis for years and has added sentiment analysis and clickstream data to see how customers interact with its website. The aim is targeting marketing with the right recommendation at the right time and optimizing plans.

On the network side, she described optimizing call routing so callers reach the right service team, planning capacity so the network is built where demand is, and predicting failures from storms or other anomalies so service can be restored quickly. "We don't want our customers to tell us that something's wrong," she said, describing a proactive stance.

She also mentioned GIS location data monetization, such as knowing where customers are so a nearby store can offer a sale.

What about customer service, fraud and marketing?

Boykin said customer service is a strong fit for AI. AT&T uses chatbots, and in interactive voice response, models help route callers properly. She said chatbots cut labor costs, are always available and can handle many tasks.

On fraud, she said telecoms face misuse of phones and fraud against the product itself, and that AT&T and other companies use AI to detect it. Predictive maintenance helps route trucks and keep vehicles and equipment working. In marketing and sales, which she said is her specialty after 11 years in marketing, AI supports customer insight and "next best action": a representative sees a screen showing a caller's past issues and products, plus suggested offers. She added that security is "a given," and that algorithms detect anomalies and potential threats.

Boykin listed example model families used at AT&T, including network quality, voluntary wireless churn, port-outs, channel preference, early technology adopters, vulnerable-care calls and broadband.

What did she say about cloud, access and chatbots?

In the Q&A, Boykin said her organization has moved to Snowflake and Azure services, with some AWS in other teams, and that it is exploring generative AI. She said she had not spent much time on how AI could improve access for low-income groups, but described a project mapping data about underserved groups back to network build plans to predict where to build out. She said AT&T has programs for low-income users, and she did not say whether those use AI.

Asked about rigid chatbots, she said older bots answered only a set number of predefined questions, while bots combined with AI build on their knowledge as people ask more questions. Asked whether AT&T uses a private LLM, she said she assumed so but that it was not her area.

How do data scientists and ML models succeed in business?

On the relationship between data scientists and ML operations, Boykin said models are only as good as the organization's ability to implement them: until people use them in day-to-day processes, "those models are just nice to have." Data scientists, she said, need to understand the business so they build models that can be implemented.

How can career changers break into data science?

An audience member who is making data science a second career asked for advice. Boykin said she asked her organization what people needed, turned those needs into data science projects and worked them on the side. A project she made up around retail, with interviews and friends' help, grew until her boss asked her to hand it to the chief data office. Her advice: "Do not wait for someone to give you a project."

What is next from AI Connect?

AI Connect continues to host practitioner talks like this one. The 2027 schedule includes 24 free live webinars, listed on the webinars page. COLLIDE 2027, the largest AI conference in the Southern U.S., takes place Wednesday, October 13, 2027, at the Sandy Springs Performing Arts Center in Atlanta; see COLLIDE. Registration is coming soon.

Frequently asked questions

How do telecom companies use AI?

Boykin described churn and sentiment analysis, call routing, capacity planning, failure prediction, fraud detection, chatbots, predictive maintenance and next-best-action sales prompts.

What cloud technology did AT&T's analytics organization use?

She said her organization moved to Snowflake and Azure services, with some AWS in other teams, at the time of the Data Science Innovators Conference 2023 talk.

How can chatbots avoid giving rigid answers?

She said combining chatbots with AI lets them build on their knowledge as people ask more questions, rather than relying on predefined problems.

How can a career changer break into data science projects?

She advised not waiting for someone to hand you a project: find a business need, work it on the side and show results.

This article is based on remarks by Dr. Frances Boykin, Director, Advanced Analytics, AT&T at Data Science Innovators Conference 2023 on April 12, 2023. More coverage: More AI Connect Events.

Filed under: AI Use Cases & Applications, Security & Fraud