First Contact

There is a long list of metrics insurers commonly use to measure the performance of the contact center: agent utilization; full-time employees per X number of customers or policies or Y amount of revenue; abandoned call rates; hold times; and speed to answer. But king among the metrics is the first-contact resolution rate.

"First-call resolution is most important because it measures the level of service you provide, not just how fast you answer a call," says Alton Martin, CEO of the Customer Operations Performance Center, Inc. (COPC), which provides contact center benchmarking.

But as the saying goes, there are three types of lies: lies, damn lies, and statistics. That's not to say traditional call center performance measurements aren't accurate, just that they might not tell the whole story. For instance, a call that is deemed "resolved," particularly when this judgment is made by the call center rep, actually may have resulted in an unsatisfied customer or caused the rep to make a bad business decision just to placate an upset caller.

Average speed of answer–a metric frequently touted by call center outsourcers–also can mask the true story. "If that's a monthly average, it may have no relevance to the experience customers have," Svoboda explains. "For instance, your average answer time could be 20 seconds. However, during peak times, your service level could be horrible, but you win it back through the rest of the week when nobody calls."

In some cases, more granular measurement can help, such as looking at average answer time in 15-minute increments. Also, metrics must be combined with analysis to complete the picture. Those analytics need to take place on both an aggregate and individual-call level. On the aggregate level, "we're seeing performance management tools, such as Performix [from NICE Systems], Merced, and Enkata, doing a better job of collecting data from multiple sources and systems, rolling up that data, and visualizing that information via dashboards at any level–agent, team, queue, or line of business," Martin says.

These types of visualization tools have enabled contact centers to do a better job of load balancing and improving performance in real time. What has been more difficult, however, is improving the quality of response to individual callers in a near-real- or real-time environment. Although insurers routinely have recorded all incoming calls and spot-audited recordings for years, the results typically have been used for training purposes.

In recent years, quality monitoring systems have added real-time speech analysis as a way to improve contact center service. "That's one of the newer twists on quality monitoring software," Svoboda reports. "A couple of vendors now offer the ability to look for tones, inflections, or key phrases."

Wisconsin Physicians Services (WPS) has been using recording and monitoring software from NICE since 2000. In 2003, WPS became a beta tester for NICE's new emotion detection and word-spotting capabilities, which then were included in a rollout of the NICE Perform platform at WPS in 2005.

NICE Perform takes a 10-second baseline reading of all customer calls. As the call progresses, the software detects increased emotion of either the caller or call center agent based on changes in pitch and tone. Calls that register high on the emotion detection scale or contain a key word, such as "hate" or "unhappy," are sent to a queue that is reviewed by WPS's contact center supervisory staff. This enables supervisors to identify calls that were indeed problems and to follow up with call center agents and customers, often within minutes.

A manual review of flagged calls is essential, says Barb Bleiler, manager of client support at WPS, because of the number of false positives that can arise. She illustrates this with an example of a call that might occur in January between a WPS call center agent, located in Wisconsin, and a caller from Florida.

"If we're in a blizzard here but it's 80 degrees and sunny in Florida, the agent and caller might have an animated discussion about the weather," she explains. "Or the agent might say, 'I hate snow.'" Although the word-matching engine can be set to tag phrases, such as "I'm unhappy about my service," doing so would flag calls with only that specific phrase, so WPS prefers to err on the side of caution.

Bleiler reports responding to dissatisfied callers quickly has improved customer satisfaction. "They've been appreciative because we've been able to correct instances when incorrect information was provided to callers, and even when the information provided to customers was correct, having someone contact them again shows we are concerned."

As an ancillary benefit outside the call center, the system has enabled WPS to check the effectiveness of marketing campaigns by monitoring how often customers mention items from promotional materials when they call. The system also can flag when customers mention the names of other health insurers. "If customers say, 'I'm going to go with XYZ company because it has a promotion,' that tells us what competitors are doing," Bleiler points out.

Replacing switch-based automated call distribution (ACD) with VoIP has been the technology development that has most significantly impacted the performance of the call center in recent years.

"VoIP allows people in distributed groups and offices to work as one unit without creating complicated network routing rules," Svoboda says. "It's more seamless for the customer and much simpler for the insurers."

From a performance monitoring stand-point, Martin issues one caution for insurers switching from ACD to VoIP. "If they choose a different vendor for VoIP than they had for the ACD switch, they may not get the reporting they used to, and people suddenly lose visibility into what's going on in the call center," Martin notes. "VoIP is definitely here to stay, but we don't see VoIP cutovers without pain."

HealthNow New York can attest to experiencing that pain. "Reporting was a huge challenge," says Daren Springer, vice president of the insurer's corporate claims operations. "Unfortunately, our initial choice for our VoIP vendor was not the best. Today, we use Cisco's VoIP platform and its ICM [Intelligent Contact Management] call manager and have been able to work through some of those problems."

Implementing VoIP from Avaya in early 2007 enabled Unitrin Direct to create a single virtual call center comprised of two physical offices in Pennsylvania and California. "Previously, we used BSR [best service routing]. Our telecom system looked to see which office had the shortest wait time and would send calls to that location, but that was no guarantee to the caller the answer time would be shorter," says Rob Shapard, assistant vice president of operations at Unitrin Direct. "Now, if you are the next caller, you will get the next available rep. We don't need to call balance [between offices] because everyone is logged into the same queue."

Unitrin Direct measures the hard-dollar benefits of the change to VoIP in terms of its impact on companywide telecom expenses, such as reducing long-distance charges of between-office calls by 55 percent and cutting hardware maintenance costs by 40 percent due to the decrease in the amount of hardware required in each of its offices. From a disaster recovery standpoint, if one of its offices loses network connectivity, Unitrin Direct immediately can redirect call traffic without interrupting callers.

In the call center itself, hold times have decreased, although Shapard could not quote specific numbers. He says there also has been a positive impact on human resources because of "better load balancing that reduced idle time and the belief that the routing of the VoIP system is more 'fair' than the previous system."

In addition, Unitrin Direct recently installed a system from Virtual Hold to improve service during peak volumes. When a call is received and the expected hold time exceeds two minutes, the caller is given the option to remain on hold or leave a message for call back. Leaving a message still maintains the caller's place in the hold queue, and the system places an outbound call to customers when their turn comes up.

"We've heard nothing but positive feedback from customers on that system because they don't have to sit with a phone to their ear," says Shapard.

Unitrin Direct would have been required to hire nine additional call center agents to handle peak call volumes without the new system. "We were able to recoup the cost of the system easily by not having to hire more staff, and we're not losing anything from a customer experience standpoint," Shapard says.

Another infrastructure technology that has been on the radar of insurers is natural-language voice recognition, also known as automated speech recognition (ASR) or conversational voice recognition (CVR). For years, voice recognition has been touted as a way to address the limitations of keypad-based or yes/no spoken interactive voice response (IVR). However, ASR hasn't lived up to its promise because, while the ability of systems to recognize words is high, it is difficult for technology to extract meaning and context from conversation.

"We haven't seen that much success with ASR because you still don't have good natural-language questioning," Martin says. "We've also seen ASR shut down because of complaints. People simply opt out too quickly because they are frustrated with the system."

That doesn't mean insurers are giving up on the technology. HealthNow is considering voice recognition as part of a pending CTI (computer telephony integration) implementation. "Not having the member remember numbers and codes for button pushing will be the big benefit to CVR," Springer says. "We started exploring it a couple of years ago but had concerns. The key is to find a good CVR system. You only get one or two opportunities when members use it; if it's cumbersome and they get stuck, you've lost them."

An option that shows some promise is agent-assisted ASR. In this method, callers answer open-ended questions ("What do you want to do today?") with natural-language responses ("I want to make a change to my automobile policy"). If the system cannot determine the meaning of what a caller says, a recording of the phrase immediately is streamed to a call center agent who navigates the IVR to the correct destination, unbeknownst to the caller.

"Humans are very good at understanding intent, so this technology pairs that capability with the efficiency of IVR," Svoboda says.

Despite the advances in technology specific to the call center, a more fundamental IT challenge exists for insurers across all verticals. "Fragmentation is a huge problem most companies are fighting with because of the number of back-end systems they have," Svoboda says.

Not only does this make it difficult to get a complete view of the customer across multiple claims, policy, billing, and other systems, but it also can lead to impersonal service. "Service reps ask, 'What's your policy number?' and not, 'What is your name?' because of the way the systems they deal with are constructed and disconnected," Svoboda points out.

HealthNow is tackling the siloed-system challenge in its ongoing effort to optimize its service center. "As we worked with our CSRs, they identified they could make a lot of improvements in how they serviced the member from an information perspective," says Springer. "They were having to go to a number of different systems to handle member calls, and sometimes the information was different between those systems."

HealthNow believed while creating a single customer information file and building a data warehouse was the best long-term solution to that problem, that strategy didn't address more immediate service problems. "We needed a quicker hit," Springer says.

In 2006, HealthNow deployed a new CRM front end, based on IntelliSource Healthcare Solutions' Streamline call center platform, to consolidate information from different back-end systems for the majority of customer requests the insurer's call center receives.

"The number of systems we needed to interface with was the biggest challenge to the project, including making sure we pulled together information on all the riders a plan had so that members were getting accurate information," says Lynn Sherk, CRM project manager at HealthNow. "Some were proprietary systems we could integrate with more easily upfront. Those we couldn't, we've put a link to within Streamline until we can [integrate them]."

"Our objective was the 80/20 rule–providing all the information in one place that agents need to service 80 percent of the calls they receive," Springer says. "We had to draw a line in the sand and get it deployed, or we'd still be working on it." Typically, HealthNow rolls out monthly upgrades to Streamline with new functionality and additional integrated information. To date, the strategy has enabled the carrier to reduce average call handle time, CSR training time, and inquiry accuracy, each by about 10 percent.

One of the next steps HealthNow will take to further reduce call handling time is to provide more complete information about callers to call center agents. "Our time studies show it can take between 20 and 30 seconds to verify a member and employer or group. Using CTI, we will be able to identify the caller upfront in an automated fashion. The sticking point to deploying it really has been HIPAA and privacy, to make sure those rules are followed and we are qualifying people appropriately," Springer says.

CTI and CRM front ends are both established and important technologies to the call center, but there is one technology commonly missing in insurers' service optimization strategy: an effective, up-to-date knowledge base. "We hear call center managers talk about knowledge bases and how important they are, but if you ask them what they've done, oftentimes it's nothing," Svoboda says.

The problem with knowledge bases is not the technology, Svoboda adds. "They fail because the content isn't maintained. IT doesn't know the answers [to business questions], but if it's up to the business to maintain the knowledge base, it doesn't know enough about the technology to create something that's easy to use and well organized."

Although technology is indispensable in maximizing the performance of the contact center and metrics are crucial for monitoring it, insurers also should heed the indirect and anecdotal measurements of their call centers. "Agent attrition gives you great insight," indicates Svoboda. "If [call center agents] are leaving, they are unhappy, and it's usually because they can't service the customer effectively. It's a people business, and your people can tell you more than technology can at times."

"Ultimately, quality service is not what we think it is; it's what the customer says it is," Bleiler says. "That's why we take the time to listen to customers–not just to hear them but really to listen to what they say."