MIT Technology Review

Wednesday, March 14, 2012

How Your Mobile Carrier Decides Whether to Pamper You

Carriers are turning to sophisticated data analytics to give better service to high-paying customers.

The controversy over "data throttling" has drawn attention to the way mobile carriers manage their networks: they target heavy users—the ones with unlimited data plans—by slowing down data-transfer speeds after they hit a certain threshold.

But carriers worldwide are also using sophisticated data analysis to give other customers—the ones with high-priced plans—gold-plated customer service to keep them happy, industry players say.

Carriers increasingly have access to huge quantities of data about their users, and they are mining this information for all sorts of insights. This is part of a broader trend across many industries; the practice of collecting and analyzing this information is often referred to as "big data."

Carriers use the big data they collect to detect whether customers enrolled in profitable data plans are experiencing problems or complaining. If so, the carriers then pamper those people selectively, says Chris Lynch, CEO of Vertica, a Cambridge, Massachusetts, firm that provides data analytics to many mobile carriers, and which was recently purchased by HP.

Carriers merge data about network problems—such as how many dropped calls a consumer experienced—with unstructured data such as the transcripts of complaints to customer-service representatives, deciphered by voice-recognition software and searched for angry keywords, Lynch says.

For customers enrolled in expensive "tiered" data plans, the carriers are vigilant to respond with refunds or discounts on re├źnrollments; they tend not to be so generous to customers with resource-guzzling unlimited data plans, Lynch says.

Lynch and other industry officials discussed such mobile-data-mining trends at a conference in Cambridge, Massachusetts, today.

Lynch did not name carriers, but says selective customer-service treatment is common industry practice in the United States. Representatives for the main industry trade group, the Communications and Telecommunications Industry Association, and from AT&T declined to comment.

Last month, AT&T decided to slow down data-transfer speeds to unlimited-data customers once they'd consumed three gigabytes of data in a month. This was an adjustment from a previous policy in which the carrier said the top 5 percent of data users with unlimited plans could be subject to throttling.

In Europe, the disparate treatment of users on different plans goes beyond offering better customer service to high-paying customers, Lynch says. There, when congestion forces a carrier to push users to another carrier's network, customers on unlimited data plans are the first to be pushed out.

"If I have a premium plan, [European carriers] want me on their network and to make sure I have the most robust, reliable experience," Lynch says. "I can identify in real-time—as the switching infrastructure is making decisions—who I want to be on and off network."

Since the carrier in the second network will offer the best service to its own customers, users that have been handed off from another carrier will bear the brunt of any congestion-related slowdowns. 

The telecom industry is working hard to sharpen the analytics tools its uses to tweak rates, and to refine service quality in specific high-traffic areas, says Chetan Sharma, a wireless analyst in Seattle. "They will use it for all sorts of purposes—they will try to monetize what can be understood from the big data," he says. "It's about how they want to treat congestion and how they want to treat customers who fall in different tiers and at different points in time."

In the United States, there is no indication that carriers give people with unlimited data plans subtly inferior treatment, except in imposing throttling, as the carriers already spell out in their policies.

Any such disparate treatment would not run afoul of FCC network neutrality regulations, which only apply to wired connections. And in any event, anecdotal reports about inferior service are hard to ascribe to any cause, because many factors can contribute to slower access or bumpy downloads.

As part of this trend, Nokia Siemens Networks launched a new tool last month at the Mobile World Congress in Barcelona that lets carriers mine and view information about network performance to find ways to change users' rates or improve service. For example, if a client used less data, the carrier could tell whether service in that neighborhood was poor and whether it would be profitable to improve it by installing another transmitter.

Of course, carriers also have troves of data on past network usage and Web access—information that will likely be used to sharpen targeted advertising in the future. "We're at the beginning of an era in big-data analytics," says Antonio Rodriguez, a venture capitalist who works at Matrix Partners in Cambridge, Massachusetts. "If you think about the treasure trove of data they have—it's question of how they tow the privacy line between what they have access to and what they do with it."