Printer Friendly

The 2016 implementation awards.

Speech technology is only as good as the outcomes it delivers. When it comes to results, our four Implementation Award winners didn't disappoint. This year, we recognize deployments that have enabled consumers to interact with their TVs using voice commands; reduced unproductive time by 800 hours per month at a consumer banking company; increased reimbursement by $23 million at a hospital; and improved conversion rates by ensuring script adherence at an online English language school. While all four Implementation Award winners come from different industries and yield different results, their deployments suggest one common trend-speech technology's value to organizations and end users continues to climb.--By the editors of Speech Technology magazine

customer: Comcast

vendor: Comcast

product: Comcast Voice Remote

Cable television provider Comcast wants to free TV viewers from painstakingly clicking their way through the alphabet to spell out, say, "NFL" or "The Big Bang Theory" on search screens; instead, it wants to make the cable-viewing experience easier by letting you voice your preferences.

And it's been quite successful, with the May 2015 launch of the Xfinity remote with voice control, which allows users of the X1 operating system to search for and navigate tens of thousands of shows by speaking into the remote.

The cable provider's X1 platform is billed as personalized TV. It brings cloud-computing technology to Comcast's technology network, so users can view the same customized screen--with apps, social media features, and traditional video services--on their tablets, their smartphones, and the TV itself. Viewers with the X1 operating system can search through live TV, on-demand shows, and their own DVR library for the shows they want to watch.

Comcast released X1 in March 2012; January 2013 saw the debut of an iTunes-available voice app meant to serve as the X1 remote, says Jeanine Heck, Comcast's senior director of product development.

"We got a really great response from people using it, but it was tough to get engagement on an app," she says. "We knew from the outset this thing would be a lot more useful if we put voice recognition on a typical, physical remote."

Viewers' TV show searches were slowed because they had to pick up the phone, open it, and launch the app, she says. Also, the smartphone-housed app introduced friction and noise into the remote, which also slowed down searches.

Comcast developers began work on a voice-activated physical remote and were able to fine-tune and optimize its features by observing the app users' habits, Heck says.

The X1 remote--the physical version, that is--launched in May 2015 and is now delivered with every new X1 installation. Though about half of Comcast's 27.7 million customers are X1 users, Comcast estimates about 8 million remotes are currently active in customers' homes, Heck says.

So how does it work? Simply, Heck says.

"You just press a button and push it to talk," she says. "So you say 'ESPN' and it changes the channel to ESPN. Or you say 'Match, Modern Family' and it brings up all the episodes of Modern Family."

Pause, rewind, fast-forward, and other features are also voice-activated, Heck says.

"Or you could say, 'Show me all the movies on HBO,' or search by actor or sports player. There are hundreds of ways you can use it," she adds.

Comcast developed the remote's hardware and much of its software in-house but teamed with a "big name" automatic speech recognition company to provide the speech-to-text capabilities, Heck says.

Feedback has been positive, she adds.

The company's analytics software shows there are 200 million voice commands executed each month, with each customer averaging about 11 commands per week. Heck notes that Comcast tracks user numbers only in aggregate, not individually.

The technology has also increased X1 viewership. The operating platform saw a 25 percent increase in on-demand hours watched; a 20 percent increase in video-on-demand revenue (Comcast doesn't make revenue numbers public); and a 30 percent reduction in the voluntary churn rate since the physical remote was introduced more than one year ago, Heck says.

Since the remote's release, transactional revenue from X1 Video On Demand customers has been 3.7 times higher than that of national pay TV customers of video on-demand services, she adds.

The cable provider has a new user goal.

"Our goal is to get X1 to half of our Comcast view footprint by the Olympics," she says.

That means she wants the remote in the hands of more than 13.5 million customers, a 5 million increase in the number of current remote users. Can it be done? Well, the feat isn't Olympian. Heck gives the gold-medal move a big yes.

--Jean Thilmany

the results

After launching its X1 voice remote in May 2015, Comcast has seen:

* 8 million remotes active in users' homes;

* a 20 percent increase in revenue from video on demand; and

* a 30 percent reduction in the number of users turning off the X1 service.

customer: Encore Capital Group

vendor: CallMiner

product: CallMiner Eureka and MyEureka

CallMiner's speech analytics is helping Encore Capital significantly improve its engagement and conversion rates through the predictive analysis of previous interactions and its behavioral impact on account managers, says Rick Britt, director of consumer data management and strategy at Encore Capital, the world's largest debt buyer. More than 1,000 account managers located in the United States, Costa Rica, and India make up to 100,000 outbound calls per day to work with consumers in resolving their debt issues across the U.S.

Organizations frequently rely on speech analytics solutions such as CallMiner for compliancy reasons. Encore's focus with CallMiner, however, is on using data from prior interactions to modify how agents speak and direct their next action.

Debt recovery must adhere to strict regulations such as disclosing the nature of the call at the beginning of each interaction. Encore found that the most unexpected benefit of CallMiner was the ability to analyze how account managers were delivering their "mini Miranda" intro including using specific words, the order of delivery, and even the impact of accents. CallMiner makes it reasonable to review the performance of all agents within these intros.

Encore discovered a remarkably tight correlation between better delivery and dramatically reduced hang-up rates within the introduction. As a result, account managers have been trained using the evidence generated by CallMiner to modify how they speak, reducing hang-up rates to an average of 17 percent from figures that were climbing as high as an alarming 45 percent.

CallMiner enables Encore to put to work the tremendous volume of metadata it has gathered from previous dialogues. More than 70 fields are utilized within CallMiner to generate predictive guidance for account management prioritizing, along with words to use to make conversations with debtors more productive. Sentiment, vocabulary, and other conversational elements are correlated with the use of CallMiner, plus Encore programming, to restack workloads, shift assignments to specially skilled agents, and prioritize calling. Results have included a 72 percent reduction in unproductive time, with a corresponding revenue increase of $1.4 million.

CallMiner's contribution toward positive business results across a number of areas stems from the ability to make personalization a business benefit. Previously, Encore considered each account as "data," with statistics derived from credit reports providing some input on likelihood to pay. With CallMiner, Encore can now overlay the dimension of "who that person is" based on characteristics filtered by CallMiner's speech analytics to make conversations more effective. As Encore notes, "This data is unpurchaseable." As a result, conversion rates have gone up more than 7 percent, with a revenue increase measured in the millions of dollars.

Personalization benefits are touching the lives of agents as well. Account managers can now listen to what are categorized as their good calls or bad calls on demand, helping them to self-modify their performance. Also, speech analytics is being used to train agents on specific issues.

Encore references a group of account managers who were having difficulty with language skills. Managers were able to identify words marked as frequent problems within CallMiner reports; they were then able to change scripts to avoid the most problematic words and phrases and used an internal wiki to help modify the delivery of other conversational elements.

CallMiner has enabled management to more effectively assist their account managers via the use of evidence that is highly personalized to each agent. Speech analytics is not being used as a "compliance hammer" at Encore; rather, account managers and their supervisors are taking advantage of CallMiner's resources to enhance their debt collection results and advance within the organization--some account managers have been promoted due to recognized improvement.

But it's not just about the technology. "CallMiner was the calm, we were the storm," Britt says, describing how CallMiner contributed to Encore customization requirements. Encore has taken CallMiner's speech analytics foundation to new heights with significant integration to internal systems and data resources, supported by close collaboration with CallMiner's technical team.

Encore Capital is one of the rare organizations that can say that "our best customer never comes back." Many are clearly not returning as CallMiner helps Encore Capital make speech analytics the center of Encore's operational intelligence.

--Steve Chirokas

the results

After implementing Eureka and MyEureka, Encore Capital Group has seen:

* a drop in unproductive time of 800 hours a month, a 72 percent reduction;

* a $1.4 million revenue increase, attributed to increased productive time;

* a 7.83 percent increase in conversion rates, with significant revenue gain; and

* reduced account manager attrition with higher satisfaction.

customer: Florida Hospital

vendor: Nuance Communications

product: Nuance CDI

Recent healthcare provider changes made it clear to officials at Florida Hospital that "we're paid for the services we document, rather than the services we provide," says Jeff Hurst, senior vice president at the nonprofit healthcare organization, based in Orlando, Fla.

To aid documentation, the nonprofit Florida Hospital launched the Nuance Clinical Documentation Improvement (CDI), which uses the J.A. Thomas Compliant Documentation Management Program. The hospital system phased in the software over a 10-month period, through May 2015.

"We launched CDI because it became clear to us that healthcare is changing from a fee-for-service to a fee-for-value model, from a financial standpoint," Hurst says.

The hospital system now uses the patient documentation software throughout eight facilities in central Florida metro locations that together comprise 2,600 hospital beds. Florida Hospital/Adventist Health System operates 25 hospitals across Florida.

The hospital system is next phasing in computer-assisted physician documentation technology, also from Nuance, that brings language understanding and voice recognition into CDI software. It calls upon Dragon Naturally Speaking for dictation and for access and clarifications to customized CDI terms.

The current Nuance program used at Florida Hospital is much the same as the language understanding solution in that it marries clinical, documentation, and coding processes. It translates a patient's clinical status into coded data that is used for quality reporting, physician report cards, reimbursement, public health data, and disease tracking and trending.

With the new system in place, Florida Hospital doctors and nurses document patients' cases from the moment they enter the hospital until final discharge and even beyond, should they see another Florida Hospital provider or be readmitted.

The system ensures documentation language is compliant with standards and that the complexity and severity of a patient case is properly noted. The latter documentation becomes part of what's called a case mix index (CMI).

The CMI ensures appropriate reimbursement because the greater the number of complex and severe diagnoses a healthcare system manages and appropriately documents, the higher its CMI. Improved documentation can raise a CMI number and, with it, reimbursement, Hurst says.

The move to CDI has already surpassed expectations, he adds. The hospital system has already seen significant improvements with reported CMI numbers. Since deployment, Florida Hospital has seen a 29 basis point increase to its CMI, from 161 to 189. That improvement translates to a $72.5 million reimbursement increase since implementation.

Documentation has also raised the hospital system's managed Medicare by 17 basis points--a measurement of Medicare patients seen with complex diagnoses--for another $23 million reimbursement increase, Hurst adds.

"We've always provided high-quality care to patients, but we just haven't always documented that," Hurst says. "Now we're getting credit for it."

Also, the observed-to-expected mortality rate dropped by 48 percent following CDI implementation. That's a nearly 50 percent decline, not because fewer patients died in the hospitals but because observed mortality rates can be artificially skewed if risk of death isn't documented or if a patient is readmitted with a life-threatening illness, Hurst says. The new system notes death risks.

With the implementation, Florida Hospital now complies with ICD-10, a clinical cataloging system that went into effect for the U.S. healthcare industry in October 2015.

The savings come with the Nuance system, even though the hospital has brought on 40 registered nurses to help physicians use and get up to speed on the CDI system.

"They're having real-time discussions with the physicians, saying things like 'You said this, but did you really mean that?'" Hurst says.

But the nurses do more than clarify physician intent. They also identify gaps in the clinical evidence and documentation and ensure documents are compliant with documentation rules and regulations. Their aid helps maintain timely documentation. The nurses are also familiar with privacy, security, and confidentiality rules that affect clinical document and information sharing.

Florida Hospital plans to next roll out the system to its healthcare clinics.

With those increased reimbursement numbers to back him up, Hurst doesn't equivocate in his praise for the CDI system.

"CDI was the shining star of 2015 and we expect even greater outcomes as we expand into the patient care and ambulatory settings," he says.--Jean Thilmany

the results

After implementing Nuance CDI, Florida Hospital has seen:

* a $72.5 million reimbursement increase;

* a 48 percent drop in the observed-to-expected mortality rate;

* a $23 million managed Medicare reimbursement increase; and

* ICD-10 compliance.

customer: Open English

vendor: CallMiner

product: CallMiner Eureka 9.4

Open English is a Miami-based organization that teaches English via online live classes with groups of students and private lessons. Since 2007, Open English has helped more than 400,000 students achieve English fluency in Latin America and beyond.

Open English contact center agents located in Colombia, Brazil, and the United States enroll new students by outbound dialing to prospects who have expressed interest though a website or campaign inquiry. Increasing enrollment has made the ability to monitor agent performance or gain insight from calls challenging, as manual efforts limit the volume of calls that can be reviewed.

Over a year ago Open English reached out to CallMiner to explore how automated monitoring using speech analytics could help improve the acquisition of new students. Open English suspected that top performing agents tended to adhere more closely to an established script. To evaluate this hypothesis, CallMiner Eureka was deployed to automatically track the progress of dialogues for more than 200 agents distributed across multiple contact centers.

CallMiner Eureka provided a framework that made it easy for Open English to break down components of interactions to consider the importance of specific words and phrases. A quality monitoring form (QMF) was established to set a mark that would have meaning for agents, supervisors, and Open English leadership. The QMF took advantage of CallMiner's ability to evaluate dialogue elements, enabling Open English to add emphasis to more important script components.

Supervisors were then able to use QMF scores as a comparison against conversion rates. What they found was a strong disparity between script followers, who performed well, and agents who tended to deviate from the script, who did not. Supervisors then used the evidence available within CallMiner for training by gathering examples from the top 10 and bottom 10 performers to modify behavior.

Agent performance consequently improved, but that was not the sole benefit. Esam Rhman, business intelligence manager for Open English, maintains that an "aha" moment occurred when it became apparent their agents were having problems handling customer objections. CallMiner provided evidence indicating that agents were not always utilizing established rebuttals for comments such as "I am not sure about the price," or "I need to talk with my spouse."

The Open English QMF score was then updated to emphasize attempting at least two of the suggested rebuttals. Conversion rates improved as QMF score monitoring and training contributed to agent success.

Efficiency and effectiveness were addressed as well by utilizing CallMiner's ability to identify silence parameters. There is always an acceptable level of silence, but excess silence could point toward areas that need improvement, such as hanging on too long with answering machines. A silence threshold was established within CallMiner to spotlight calls that were likely unproductive. Open English attributes 1,000 hours per month gained by avoiding unnecessary silence with savings of $5,000 per month. Agents were also able to make more calls as a result.

Perhaps most impactful was CallMiner's ability to help Open English link its sales conversion process to marketing efforts. Open English created an automated score within CallMiner that tracked Urchin Traffic Monitor (UTM) parameters from websites, search, social media, and other marketing campaigns. This enabled correlation between marketing and conversion success. A CallMiner application programming interface made it easy for Open English to create on-demand access to this detail for marketing. As a result, marketing has become more agile in managing its campaigns.

Open English has worked closely with CallMiner to the benefit of both organizations. CallMiner's Spanish module has addressed Open English's almost 100 percent Spanish-speaking customer base. Open English has contributed toward improvement of the Spanish module by adding to its dictionary based on dialect components gathered from a range of Spanish speaking geographies.

Future initiatives include using CallMiner to correlate interactions with conversions that occur over a period of time. Details from this metadata could be used to modify follow-up calls or ignore those that rarely result in a sale.

CallMiner has provided Open English with insight into its prospect base that was previously unattainable. Additional benefits have helped span what many consider as a frequently seen chasm between marketing and sales. As a result, Open English is able to put business intelligence to work to increase revenue and reduce costs for its entire organization.--Steve Chirokas

the results

After implementing CallMiner Eureka 9.4, Open English has seen:

* increased conversion rates by using script adherence examples to modify agent behavior;

* improved call center efficiency by examining the "percentage of silence," eliminating 1,000 wasted labor hours and equating to approximately $5,000 a month; and

* linked marketing efforts to sales conversion to improve agility.
COPYRIGHT 2016 Information Today, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Speech Technology Magazine
Date:Sep 22, 2016
Previous Article:The 2016 star performers.
Next Article:Voice biometrics can protect against fraud--and ensure customer satisfaction: secure customers--whether they're using your mobile app, your contact...

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters