Are real-time analytics essential for AI-based speech analytics solutions?

Matt Dixon

July 12, 2018

At the recent Customer Contact Week event in Las Vegas, nearly every potential customer we spoke to seemed to be very focused on a handful of features—but, when pressed, had a hard time explaining why those features were necessary to deliver business value.The clearest example of this is “real-time analytics.” What this means is that an AI-based speech analytics solution is capable of analyzing conversations in real-time—as a rep is talking to a customer. But really, the most common use case that requires this level of analytics speed: screen pops. A screen pop is when a system flashes a recommendation in front of a representative while he or she is speaking to a customer. Sometimes these prompts contain suggested scripting (“Be sure to thank the customer for her loyalty”), sometimes suggested actions (“Offer the prepaid service option”) and sometimes simple graphical prompts (e.g., a smiley face or a coffee cup) to remind reps to sound upbeat, energetic and positive.Regardless of the type of screen pop used, the goal is the same: to actively steer customer conversations toward better outcomes (whether that’s lower CES, higher NPS or sales conversion).But while the tech analyst community has taught the market to ask for it, are real-time analytics a “must-have” functionality in an AI-based speech analytics platform? What business problem does it really solve and does it do so effectively?

Beyond the screen pop

On the face of it, screen pops would seem to solve a variety of problems related to service, sales and compliance effectiveness. Companies are often challenged by the fact that reps just go through the motions when it comes to customer interactions—failing to pick up on customer cues that would enable them to reposition different options and improve the overall quality of the experience. Similarly, a shockingly high percentage of reps—even when they’re incentivized to do so—fail to “ask for the sale” or make cross-sell and upsell offers when the opportunity arises. And reps often neglect to read company-mandated disclosure statements following certain transactions.But do they work? While there are certainly a handful of examples of companies claiming success with this approach, the track record for this technology is remarkably spotty given how long it’s been in the market. The reality is that most contact centers are staffed with lower-wage and, in many cases, part-time staff, who churn at a remarkably high rate. So, it should come as little surprise that flashing things in front of reps is not a sustainable way to drive behavior change. The truth is that screen pops can be a distraction more than an assistance to reps and, over time, are tuned out and viewed by reps as little more than “white noise.”From a purely practical perspective, it’s worth asking the question of what’s more impactful to the business: alerting a rep in the moment to recover a customer who is already aggravated…or using AI to instead determine the root causes of customer aggravation and eliminating those at their source, through more effective and targeted training and coaching?But the real problem with using real-time analytics to deliver screen pops isn’t the strategy's questionable success rate or the dubious business logic underpinning the approach.The real problem is that they make things worse.

Complex environments need better solutions

About a decade ago, we reached a tipping point in the customer journey as it related to service: more than half of customers, reported first attempting to self-serve on their issues before picking up the phone to reach a live representative. Today, that number is north of 80 percent. With the “easy,” transactional issues (e.g., address changes, balance checks, simple orders, etc.) going to self-service, what ends up making it through to the live rep are the more complex, varied and one-off issues.This has necessitated a wholesale shift in how contact centers manage their people. The old “factory floor” model of agent scripting, handle time metrics and checklist-driven quality assurance has given way to a new model much more akin to a knowledge work environment in which reps are encouraged to use their judgment, harness the knowledge of their peers and collaborate to come up with solutions to complex customer issues.The research is very clear on this point: transactional models only work in simple, predictable work environments, and the mechanisms (like real-time analytics) used to manage transactional work actually have a negative impact on performance when applied in a complex work environment.

Control Quotient empowerment

CEB, now a part of Gartner, found that the single highest driver of frontline service rep performance is something they called “Control Quotient” or CQ, for short (see The Effortless Experience, a book I helped to co-author, for a more detailed discussion). CQ is an environmental factor in which reps feel a higher level of resiliency because they are more in control of their work and more empowered to use their own judgment to solve customer problems.CQ was more than 2X more powerful a driver of rep performance than any other factor studied. It has more impact than hiring reps with higher IQ or EQ. Moving from the 25th percentile to the 75th percentile on CQ drives an 11 percent improvement in rep performance. Hiring reps with higher IQ or EQ, by comparison, only drives a 4 percent and 5 percent improvement in performance, respectively.When companies hold a handle time clock over reps' heads, ask them to “stick to the script” and audit their performance using a rigid QA checklist, they're inhibiting CQ. Screen pops are little more than a modern-day script designed to wrest control from reps and supplant the judgment of a rep with the judgment of the company. When we talk to reps about this technology, they are emphatic on this point—screen pops feel like a cattle prod… a signal that the company doesn’t actually trust them to engage customers in the way they see fit.

Knowledge work over transactional work

In a follow up article that I co-authored in Harvard Business Review, titled “Kick-Ass Customer Service,” my team at CEB found that the top performing reps were a profile we called “controllers.” These are take-charge individuals who pride themselves on their authority and expertise. Controllers are dramatically more effective than empathizers, the profile that most service leaders say is the one they would look for in their next hire. What this tells us is that today’s customer doesn’t want an apology as much as they want to engage with a rep who is smarter than them regarding the issues they’re calling about.It’s worth acknowledging this does not mean ALL real-time analytics are bad. Particularly in compliance settings, real-time analytics around fraud detection could be very valuable. Similarly, when managers are able to quickly spot problems regarding critical business issues they can pivot in meaningful and timely ways. But most managers we work with tell us in those cases, within a few hours is plenty fast enough. So, it’s less about the speed (in the moment delivery) and more about how and by whom it is used.As appealing as real-time analytics and screen pops seem to be, they are little more than a modern twist on an outdated management philosophy. Our strong advice is that leaders spend their time and resources on technologies that enable knowledge work, not technologies that reinforce transactional work.

Most popular articles