AT&T is eavesdropping on its disgruntled customers via their Twitter messages. The project is an experiment designed to automatically pinpoint where and when people are having problems with their wireless connection. Software developed by AT&T researchers finds complaints about network problems on the social network (which now has 175 million users) and extracts the approximate time the tweet was sent and the location of its sender.
More and more firms are turning to Twitter in search of their customers’ voices, but usually they do this to understand how people perceive their brand or to respond to specific consumer problems. AT&T’s project is a novel way to mine the collective mood of the tweetosphere.
When AT&T secured exclusive rights to distribute the iPhone in 2007, the deal proved to be both a blessing and a curse. The much-desired handset brought huge customer growth and brand prestige. But along with that came soaring demand for wireless data that overwhelmed AT&T’s network in places such as New York and San Francisco, leading to dropped calls and sluggish connections.
The company has an automated network monitoring system that can detect connectivity problems, and customers can, of course, call in to report problems. But by mining messages shared on Twitter, AT&T gets extra real-time information and can prioritize fixes, says Jia Wang, a member of the company’s Internet and Systems Networking Research Center. “We are trying to identify three pieces of information: where the customer experienced problems, what type of problem, and when they experienced it,” she says.
Wang and colleagues use two levels of filtering to find tweets by frustrated customers, and they do this by tapping into the programming interface tools Twitter makes freely available. A general set of queries pulls in every tweet related to AT&T’s mobile service before a more rigorous set of rules homes in on those relating to service quality, for example messages containing words like “call dropped” or “3G.” This automated method was around 90 percent accurate at identifying genuine complaints, the researchers found.
The company can use the time stamps and location data associated with collected tweets to estimate where and when a person experienced a problem, making it possible to put the records on a map. This information can be compared with system logs and customer service calls. In practice, the technique picked up on issues that wouldn’t otherwise have been reported. It also detected issues earlier—around 20 minutes earlier, on average, than a customer service call came in.
“It tells us what matters to people and what affects how people feel about our network,” says Jennifer Yates, an executive director at AT&T labs who is overseeing research on new ways to manage the firm’s network. Most grumbles on Twitter are about dropped calls, followed by slow Web speeds or lack of service. Users tend to call in only with more serious problems, says Yates.
“We are working on how we can operationalize this research so that it can be used by our network managers alongside our existing monitoring tools,” says Yates. The data will accompany that from an iPhone app, Mark the Spot, that customers are encouraged to use to report problems, she says, which is used to plan network upgrades like new cellphone towers.
Yates adds that in time, users may figure out that AT&T is listening for their tweets, and become more likely to share their experiences. “I hope they would be pleased we are taking an interest,” she says.
Michelle DeHass of Attensity, a company that makes software to track and respond to tweeted complaints, says it is unusual for a company to gauge the performance of its own systems using Twitter. But she also points out that AT&T may need better tools for handling the ambiguity of language used in tweets. “Using a list of keywords is not really enough,” DeHass says. “To find everything, you need to use semantic technology able to do deep parsing and look at families of words and synonyms,” which Attensity’s software has, she adds.
Yates says AT&T has smarter forms of text analysis in the works. “It will help understand the short language in tweets and also match them up with our other sources of data, such as customer service reports,” she says.
Capitalizing on machine learning with collaborative, structured enterprise tooling teams
Machine learning advances require an evolution of processes, tooling, and operations.
The Download: how to fight pandemics, and a top scientist turned-advisor
Plus: Humane's Ai Pin has been unveiled
The race to destroy PFAS, the forever chemicals
Scientists are showing these damaging compounds can be beat.
How scientists are being squeezed to take sides in the conflict between Israel and Palestine
Tensions over the war are flaring on social media—with real-life ramifications.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.