Nervous System: Dialing for Dollars

David Kalat looks back at how predictive dialing software revolutionized the call center industry—and left behind a legacy that continues to be litigated to this day.

Historically, call center managers struggled to identify for the best time to initiate a new dialing attempt while an agent was still occupied such that the next successful connection came as close as possible to when that agent became free. Finding that sweet spot was beyond human ability, due to the array of different factors involved, and the way those factors may fluctuate over time.

Like many call centers in the late 1980s, call center managers at International Telesystems Corporation (ITC) relied on a fixed time interval between dialing attempts. Calculating that interval as best as humanly possible still meant that call agents sat idle as much as twenty-five minutes of every hour, and that five to ten percent of the outbound calls were simply abandoned after being connected. It was a poor compromise. A Virginia data analyst and statistician, Douglas Samuelson, developed a software solution to this problem.

Samuelson’s predictive dialing software was a disruptive technology that transformed the call center industry. It also unexpectedly ended up as a central battleground in numerous class action lawsuits.

In November, the Supreme Court heard telephonic oral arguments in the matter of Facebook v. Duguid on how this technology is regulated, and their eventual ruling will affect one of the largest single categories of consumer financial litigation in the country.

At issue before the Supreme Court was the question of whether a predictive dialer constitutes an “automatic telephone dialing system” (ATDS). This is a technical question with a substantive legal significance. The Telephone Consumer Protection Act (47 U.S.C. § 227, “TCPA”) was passed in 1991 to protect American consumers from invasive and harassing advertising phone calls. The number of lawsuits filed under the TCPA has grown almost every year since its passage, and TCPA lawsuits account for over a quarter of consumer litigation in the United States. With statutory damages of $500 per violation—tripled if the violation is found to be willful—a class action brought under the TCPA potentially can add up to a liability of nine figures or more.

The TCPA makes it unlawful to, among other things, use an ATDS or an artificial or prerecorded voice message to make a nonemergency call without prior express consent of the called party. Establishing whether the technology used to place the call is an ATDS is a foundational question that in many instances can either create or dissipate TCPA liability.

The statute defines an ATDS as “equipment which has the capacity—(A) to store or produce telephone numbers to be called, using a random or sequential number generator; and (B) to dial such numbers.” That phrase has been analyzed, parsed, and dissected by litigants and courts for almost as many years as there are words in it. What kind of TCPA litigation the future may hold will depend a great deal on how the Supreme Court chooses to interpret those twenty-six words.

Autodialing technology dates back at least to 1976, when a patent was issued to Norman Sheldon and Erwin Dosch for a machine that generated telephone numbers using a sequential number generator, dialed those numbers automatically, and then played a prerecorded message. Autodialers that have the capacity to generate random and/or sequential telephone numbers therefore are able to call numbers assigned to emergency services providers or other unlisted numbers that otherwise would not show up in calling lists, or to call all of the sequentially assigned extensions at a single organization.

“Predictive dialers” differ from traditional autodialers in that they do not generate the telephone numbers they dial, but instead draw telephone numbers from a database while executing an algorithm to pace the timing of calls. That algorithm is the defining feature of a predictive dialer. The term “predictive” refers to the use of statistical calculations to pace when calls are made to optimize the call center agent’s talk time while minimizing the chances of connected calls being disconnected due to an agent’s unavailability.

The more time those agents spend on the phone talking to people, the more value they generate. If an agent waits until he or she completes a call to dial the next one, there will be an unavoidable amount of downtime while the next call is dialed and connected and the called party is brought on the line. Not every call connects to a recipient: the line may be busy, the phone number may be disconnected, the phone may ring without answer, an answering machine might pick up. Any given dialing attempt has many possible outcomes; connecting to a recipient is only one.

Meanwhile, numerous factors affect whether the call center agent is ready to take the next call. The number of call center agents may vary as employees clock in and out; the average length of a call may vary depending on the campaign being used; the likelihood of encountering disconnected numbers may vary depending on demographics; the odds of getting a person on the line may vary at different times of the day. The upshot of this swirl of unknowns is that the new call may be answered by a consumer when no agent is available. In such instances, the connection is dropped, and the consumer is left answering a “dead” line, never to know what the call was about or from whom it came. The industry calls these “abandoned calls,” but consumers call them infuriating.

This is where Douglas Samuelson’s innovation changed the game. His predictive dialing software used sophisticated statistical models to make intelligent choices about when to dial the next number. In 1987, Samuelson’s prototype predictive dialer was first deployed. By the time his patent was approved in 1989, ITC was reportedly enjoying between 50 and 55 minutes of active agent talk time per hour, and fewer than two percent of calls were abandoned. Some estimates have since found that using predictive dialing software can increase a call center’s overall productivity by up to 300 percent.

Samuelson’s predictive dialing software revolutionized the call center industry—and left behind a legacy that continues to be litigated to this day.

The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions, position, or policy of Berkeley Research Group, LLC or its other employees and affiliates.

Read the article. (subscription required)