For most of this decade, call center automation efforts involving voice technologies have focused on speech analytics to monitor agents' interactions with customers. That is changing. <\/p>\n
New voice tools provide the ability to initiate CRM contacts through voice interfaces, such as with Google Home and Amazon Echo. They can also divine new insights about customer experiences through voice analysis of calls that supplement traditional consumer surveys, and may even replace them in the future. <\/p>\n
But these voice tools can help only to a point, as software vendors and their customers await better natural language processing (NLP) support. <\/p>\n
\"It's at a point where it's really accepted -- you talk to your car, you talk to your phone,\" said John Joseph, CEO of Scribe Software, a company based in Manchester, N.H., that focuses on CRM data integrations. His team demonstrated at the Salesforce World Tour stop in Boston in May a Salesforce-Amazon Echo implementation that enables voice-activated report generation. <\/p>\n
\"There's still a long way to go in terms of natural language [processing technology] understanding everything -- but it's remarkable where we are,\" he said. <\/p>\n
The reason voice recognition or virtual assistants are still kind of dumb -- despite massive cloud compute power that can run circles around humans -- is that they are very literal, and they often over-simplify a human's request to the point where they miss the original premise. This can lead to off-topic suggestions or some variation of \"Sorry, I don't understand.\" <\/p>\n
On top of that, humans have powers of comprehension computers don't: We instinctively understand tone and emotion, and we can hear the syntactic nuance of clichs and other local idioms (such as the interchangeable use of soda, pop, tonic or Coke). <\/p>\n
There's still a long way to go in terms of natural language understanding everything -- but it's remarkable where we are. John JosephCEO, Scribe Software <\/p>\n
Think about this statement: Police help dog bite victim. How is NLP likely to interpret that? An NLP system has to work hard to figure out what you're saying (think I scream versus ice cream) before it can tackle context and hazard a guess as to what a customer is asking. <\/p>\n
All that being said, companies hope to integrate more call center automation tools built around NLP as it improves, according to Deloitte customer operations leader Andy Haas, who co-authored a report analyzing the results of a survey of 450 call center executives earlier this year. <\/p>\n
Business leaders are considering and even experimenting with next-generation voice recognition feeding into NLP, which, in turn, feeds into analytics systems that can automate customer service through the insights the analytics glean from the conversation, Haas said. While simple, targeted tasks can be completed now, operations executives pretty much agree that adequately reliable automation technology is a long way off from digitizing customer interactions; as in, decades. <\/p>\n
\"There might be a tipping point in the future, but it's not there yet,\" Haas said. \"I don't think my clients think it's in the next five years, just like operations managers don't think interactions will be all-digital in the next five years. Will it happen in the next 20 years? Maybe.\" <\/p>\n
One possible way new voice tools could advance call center automation is through analytics to determine customer sentiment for the purpose of future sales and customer retention efforts. For decades, the post-call follow-up survey has been the main method fueling such initiatives, but voice analytics are starting to supplement surveys. <\/p>\n
There could be a point down the road where these audio mining systems replace surveys, and they could actually offer deeper insights about customer sentiment than the blunt instrument of the three-question, multiple-choice survey few take the time to fill out. <\/p>\n
Haas said his survey showed that while call center volumes are going down in general, the interactions which escalate to calls are make or break in terms of the customer's relationship to the company. Call analytics tools, therefore, will become more and more important vehicles for customer retention. <\/p>\n
\"As you apply analytics, it will make an easier ROI [for investing in the technology],\" Haas said. \"It's going to be less pure volume, but more meaningful interaction.\" <\/p>\n
Greg Hirschi, director of customer service operations at smartphone and tablet case manufacturer OtterBox, runs a 270-agent call center based in Colorado. The company regularly conducts customer surveys, which get 30% to 40% response, and the rich information they yield has led directly to eight-figure redesigns of customer experience processes, one example being warranty service, he said. Analytics can extend those insights to offer product teams feedback for future OtterBox models. <\/p>\n
\"From a consumer insight standpoint, for us, it's deeply valuable to understand how they use our products and what we can do to better design them,\" Hirschi said. \"There's a knowledge gap between perceived customer use and actual customer use, and we use voice analytics to bridge that gap.\" <\/p>\n
Terry Leahy, president and CEO of call analytics software vendor CallMiner, said he believes the old-school customer survey as a service tool should be replaced, and the funds companies invest in them would be better spent elsewhere. That being said, customer surveys will never go away he added. Call analytics can offer insight to marry with the results of surveys and to deepen a company's knowledge about its customer experience. <\/p>\n
\"We are now selling to marketing more than we ever did before, and that's where the budget for the survey usually is,\" Leahy said. \"I think it's safe to say that you'll be seeing budget for surveys being diverted [toward] a better way to understand the actual voice of the customer than a derivative of it, which is the survey ... But surveys are never going away.\" <\/p>\n
Voice-over Internet Protocol (VoIP) phones have, for years, extended call center work to employees who want to work at home. But even the old-dog VoIP technology is teaching call centers new tricks. <\/p>\n
Ryan Nichols, general manager of Zendesk Talk, said CRM systems are creating deeper and deeper VoIP integrations, such that service agents can escalate calls from channels to voice while in a customer's recording, without interruptions. This reduces call times dramatically because there's no cold-call script to launch into the discussion -- it's already going on via text, and the voice call is a continuation of that. <\/p>\n
\"Conversations don't need to come in via PSTN [public switched telephone network] anymore,\" Nichols said. \"Someone doesn't have to dial in a 1-800 number they found on the website and navigate down to an agent.\" <\/p>\n
These VoIP integrations have become so tight, he said, call centers are either no longer using traditional phone systems or they're skipping them altogether when equipping new facilities. Customer agents are the better for it because, when they can see context in the customer record, as well as the chat history, agents can perform more effective service. <\/p>\n
Nichols is watching with interest how companies such as Uber and Lyft integrate voice into their smartphone apps, as well as home voice assistants such as Amazon Echo. Still, he said, there's a long way to go before we read a lot of CRM success stories tied to voice recognition and the NLP those types of implementations require. <\/p>\n
\"The question is, what happens when people have problems?\" Nichols said, echoing what analysts have said all year: NLP is unreliable enough that the biggest challenge is understanding when and how to escalate service to better channels before losing the customer. <\/p>\n
\"How do you build a bridge between this conversation that's happening around your core service and your traditional support channels?\" <\/p>\n
Guide to buying call center speech analytics <\/p>\n
The benefits and negatives of real-time speech analytics <\/p>\n
Best and worst call analytics practices <\/p>\n
<\/p>\n
Go here to see the original: <\/p>\n
Call center automation advances, but only as far as NLP can take it - TechTarget<\/a><\/p>\n","protected":false},"excerpt":{"rendered":" For most of this decade, call center automation efforts involving voice technologies have focused on speech analytics to monitor agents' interactions with customers. That is changing. Continue reading