By Curt Gooden
Managers of modern customer contact centers and the callers they service can have conflicting agendas. In pursuit of lower costs, managers want to deflect calls from a live agent to an automated solution as much as they can. Currently that means using interactive voice response (IVR) self-service solutions robust enough to serve many, if not all, possible situations customers could be calling about, without the customer pressing zero to speak with a live agent.
For their part, customers, or those needing information only a live agent seem able to provide, have their fingers hovering over and ready to press zero to reach a representative. How can managers approach their own goal of “zero” – whether it’s zero CSR interaction or zero cost overhead – while keeping customers satisfied and informed?
For a growing number of organizations, the answer is next-generation artificial intelligence. Artificial intelligence (AI) is an automated solution gaining traction in the customer contact industry and is the culmination of several technologies that have been steadily improving IVR performance over the years. It entails utilizing natural speech recognition throughout the call and an “intelligent” brain on the back end to interpret (and in some cases, predict) the user’s responses to drive call flow and processing accordingly.
High cost of entry has slowed adoption of AI within IVRs. But as with any new technology, costs have been falling as technology becomes more pervasive and user-friendly. In turn analysts expect call center adoption, more prevalent use, and an improved, even seamless, customer experience to help spur customer approval.
The Differences Between AI and IVR: The differences between traditional IVR and AI are stark. The logic of traditional IVR technology seeks to create call paths or a map of all possible user options. Press a sequence of numbers along the decision tree, and the caller should end up with the solution, as mapped in advance.
Therein is the challenge. The IVR map is static. While complex IVR can have hundreds of pages of detailed directions and anticipated interactions, trying to account for every possible customer scenario can be challenging – hence the reason so many customers press zero to speak to a live agent.
An AI solution can help reduce or eliminate the zero option. Some basic AI solutions are layered atop existing IVR applications in an effort to expand on the logic, call mapping, and speech recognition of IVR, yet with a level of sophistication found only with higher-level intelligence. In most current and future instances, though, AI will be a stand-alone solution that will replace the traditional IVR.
Here’s how it might work in an airline reservation application. The customer is instructed to speak his or her request or issue to the automated solution. For example, “Please tell us in a few words why you’re calling today. You can say, new reservations, existing reservations, change reservations, or something else.” The AI application will turn the speech into text, analyze the request, and employ algorithms to start the caller on the call map.
Intelligent mapping anticipates customers’ requests and can more proactively guide them toward the expected path. The more intelligent the solution, the more it has been programmed with variables it can anticipate, and the higher the likelihood the caller won’t press zero.
Is AI Ready for Prime Time? Some makers have reported up to 70 percent deflection rate; whether that’s a sales promise is questionable. A more realistic expectation in highly complex environments, such as a customer seeking to change an airline reservation, could be closer to 10 percent. In the healthcare industry, for example, where customers might need to select from a variety of solutions – and the customers themselves might be older, speak a different native language, or are otherwise hindered from using the system – AI as an option is likely is further out.
As such AI is not quite ready for wide-scale implementation for a variety of other reasons. The cost of AI will keep it viable only for those larger organizations that either can afford the application or will realize benefits by slicing a few minutes off each call or a few calls from each live agent.
It also will find acceptance in certain industries before others; some may never embrace AI, at least in the near term. Those who most successfully adopt AI will be travel and hospitality. As noted, thinning margins will require the constant pursuit of solutions that shave minutes or manpower from each customer interaction. Moreover, some engagements are not extremely complex, like booking one hotel room for a weekend.
Customers in the travel and hospitality segment typically are more technologically savvy and quick to adopt new solutions. Once AI matures, the healthcare industry might realize some of the greatest benefits from AI.
Along those lines, the cost/benefit analysis of AI is a simple calculation, notwithstanding higher initial costs. If a provider is charging by the minute or call, every minute saved or call deflected translates to savings. In fact, a simple revamping of traditional IVR or an improving of the call tree can yield cost savings. With AI the chance to increase deflection can be exponential.
As technology advances, artificial intelligence will find its way into interactive voice response. Though comparatively costly now, AI can be a justifiable expense for the right organizations. In time it will become the industry standard – especially for those organizations hoping to reduce the time agents spend on calls and serving those customers who hopefully prefer to not chat with an agent at all.
Curt Gooden is the senior vice president and chief information officer at C3/CustomerContactChannels with nearly two decades of experience leading IT departments for global companies. He has successfully managed all aspects of information technology across international boundaries while lowering costs, implementing controls, mitigating risks, and adding tangible value to clients.
[From Connection Magazine – November/December 2015]