|
Facts, Figures and First-Rate Call Centers By Lee Hollman
Among the myriad of challenges that call centers face, garnering raw data about their customers and operations is not one of them. They can access mountains of call statistics from phone switches. They can, using CRM software, generate reports ad infinitum detailing what purchases customers make and customers' opinions about your service. Call monitoring tools can fill volumes with figures on agent talk time, hold time, post-call wrap-up time, etc.
The challenge for call center managers, rather, is to aggregate data in real-time from these and other sources - your ACD, workforce management software, e-mail and Web servers - to make timely and informed decisions that advance call center objectives.
Example: Analytics tools can help agents recognize a cross-selling opportunity by noticing that a high number of customers who purchase PCs over the phone later purchase memory expansion modules on-line. Your company's sales department can view the same data and instruct the phone agents to offer PC buyers a special promotion on memory expansion modules - thereby boosting revenue on the original sale.
Analytics can also help you troubleshoot customer problems. If agents receive frequent calls about a defective product, employees from various departments can view the rising call volumes from upset customers and react accordingly.
Donna Fluss, principal for DMG Consulting, says analytics opens up key customer service data to sales, marketing and other departments within your organization.
"Sales and marketing departments don't take agents' comments very seriously," she says. "But if there's a process that automates the collection of that data, analyzes that data, expresses the data's relevance and allows the organization to take action, the input from the call center will be taken more seriously."
Vendors featured in this article offer "real-time analytics," tools that allow you to share the timeliest data with agents and other employees. But the term requires clarification.
Fluss and others note that it's hard to capture, analyze and prepare the data so that your organization can take action on the fly. The majority of real-time analytics software delivers data about customers' actions within five minutes of their occurrence. All of the vendors we interviewed say their products can update data within five to 15 minutes.
That's still a sufficient amount of time for agents to receive important information, such as the likelihood that a customer might purchase a given product or cease doing business with you.
"You've got to deliver analytics to agents while the customer's on the phone," says Mike Schroeck, global iAnalytics leader for IBM Business Consulting Services. "To that extent, [analytics software] turns call centers into arms of marketing departments by enabling agents to ask the right questions to customers."
As useful as analytics can be to secure both sales and customer loyalty, the software can also help improve call center efficiency. By capturing data about agents' job performance, like how quickly and successfully they resolve each call, you can determine what aspects of customer service to enhance.
This article will focus on analytics tools that collect data from multiple sources, like phone switches, e-mail and Web servers, CRM software and databases containing customer information.
Given space constraints, we cannot describe every analytics product that generates reports containing information about agent performance or customer service trends. Many of these products have, and will continue to be, covered in separate product features (e.g., the analytics modules of call monitoring and CRM software vendors, which we described, respectively, in our October and December 2002 issues). This article will address "real-time" analytics tools from vendors that are comparatively new to our pages or otherwise do not lend themselves to other feature topics.
In addition to detailing product functionality, the vendors we interviewed detail how best to use analytics to anticipate customers' needs, increase sales and sharpen agents' customer service skills. Here are their insights and opinions.
An Assortment Of Analytics Angles
Bob Kelly, president and CEO of CenterForce Technologies (Bethesda, MD), explains why most call centers can benefit from integrating data with analytics software.
"If you're [managing] a simple center that handles inbound voice calls and the only thing you might need to know is how agents spend their time on the phone, you can get that from phone system reports," says Kelly. "But if I want to find out anything more complex, like the relationship between how agents use CRM or collections software and the number of sales dollars they earn per hour, I need analytics software."
Kelly adds that you don't need to manage a complex call center to gain a substantial ROI with analytics. If your center handles only inbound customer service calls, you can use the tools to determine the relationship between agents' call handling times and customers' satisfaction with the resolution of each call. You might find that shorter call lengths don't make for happier customers. Viewing agents' average handle time statistics from your phone switch can't help you to make such a distinction.
Analytics' effectiveness, say observers, hinges in part on figuring out what devices or sources to collect data from, and what data to collect.
"Seventy percent of the work is getting the data," says Michael Smith, group director of marketing for Informatica (Redwood City, CA). "The remaining 30% is defining what metrics you want to use."
Smith places analytics into three categories. The first is call statistics that you can find from a phone switch, like average speed of answer and call volumes. Category two comprises the quality of customer service, like how many customer service cases were opened or closed. The final category, focusing on agent performance, might detail who helped the most customers or attained the highest customer satisfaction ratings. Combining metrics from each category lets you determine agents' strengths and weaknesses.
But don't err by viewing agents' performance metrics only on an individual basis.
"The way to determine if an agent isn't performing up to standard is seeing what they've done relative to the objectives you've set and compared to other agents," says David Spindel, technology analyst for consulting firm Datamonitor (New York, NY).
He adds that analytics can help you assess how well agents perform relative to predefined objectives. If you set an hourly sales goal and note that agents consistently don't confirm as many sales as you expected, you might need to reassess that goal. If you find that the majority of agents exceed your goal, you can contemplate raising the bar.
Comparing agents' personal metrics can also help you identify and reward star performers, as well as agents who need additional training.
"Analytics by itself will not and cannot drive changes in agent behavior," says Rosemary Turley, vice president of marketing for Performix Technologies (Burlington, MA). "You need to take it one step further and make sure that the information that you gather translates into something that makes each individual in the organization a little more productive and a little more happy."
Analytics tools can also contribute to making your customers happier and your business more profitable. One common strategy is to focus on customers' purchase habits to determine what additional products they're interested in. Another is to identify trends in customer behavior, such as how often customers contact you through different media, what pages of your Web site they view and the nature of their communications with agents. Many companies combine both strategies to plan targeted marketing campaigns.
Office supply chain Staples used SAS' (Cary, NC) analytics software to study customers' spending habits. Staples found that customers who purchased items from the company's retail stores, Web site and through Staples' print catalogue bought more merchandise than customers who did their shopping through only one of those channels. By recognizing this trend, Staples encouraged customers to use more than one channel and boosted its sales.
If you use analytics to track customer behavior, be sure to pay special attention to extreme changes in customers' purchasing habits.
"Increasing purchases or decreasing them are indicators that the customer's situation is changing," says Mark Smith, president of Quadstone (Boston, MA). "It's an indication that something's going to change in a customer's life, or that they can make the decision that they've had enough [of your products or services]."
Smith isn't alone in his opinion. Kevin Cavanaugh, vice president of technology for Unica (Lincoln, MA), says that any major life event, such as a marriage or birth, can change a customer's spending habits.
Cavanaugh acknowledges that using analytics to secure customer loyalty is an "inexact science," adding that you can't expect to keep every customer. But he insists that the software can significantly strengthen customer loyalty.
"It gives you estimates of the likelihood that a customer is going to do something," he says. "And if you can retain even a few percent more customers, it has a huge value to your business."
Analyzing Your Analytics Software Options
All of the real-time products in this feature provide analytics related to multimedia communications between agents and customers. But some products have a specific functionality that might suit your business priorities.
If your focus is on boosting agent productivity, CenterForce Technologies and Performix Technologies let you share performance data with reps so they can identify their strengths and weaknesses. Products from CSG Systems (Englewood, CO), Informatica, Quadstone, SAS and Unica, by contrast, tend to emphasize maximizing sales opportunities and keeping customers.
That's not to say that you couldn't use analytics apps from any of the vendors mentioned above for both sharpening agents' customer service skills and generating revenue. But each developer's software has distinctive merits.
For example, we found that Performix Technologies offers additional modules for its analytics software to help you develop agents' skills after you've viewed statistics to assess their competence at their jobs. And Island Data (Carlsbad, CA) calculates the probability that customers will enter specific words together during a search of your on-line knowledge base to more accurately understand customers' requests.
But since all analytics software can offer valuable information about your call center, how can you choose between so many different products? Here's a quick overview of some of the options available to help you make an informed decision.
CenterForce Technologies' CenterForce Analyzer integrates with workforce management software from vendors like Blue Pumpkin, IEX, and RightForce, the last of which merged with CenterForce in November 2002. For outbound campaigns, you can use CenterForce Planner to forecast the number of agents you need during each shift. CenterForce Analyzer then compares the actions that agents are scheduled to perform, like answering calls or taking lunch breaks, to what they're actually doing.
To set criteria for evaluating agents, you can select Performance Indicators (PIs) in CenterForce Analyzer. Although the software originally included 17 Performance Indicators to rate how well agents respond to inbound calls, the latest version includes 70. The newest PIs focus on agents' abilities to handle outbound calls and non-real-time interactions (e.g., the number of outbound calls they make that end in a successful sale or the amount of time they spend answering e-mail.)
You can set goals for agents based on the Performance Indicators that you select. If you select a PI to find out how much time agents spend answering e-mail, you can set a goal requiring agents to respond to each e-mail message within one hour. The goals that you set appear alongside analytics that indicate if agents meet them on scorecards that you can e-mail to individual agents and groups of agents.
The latest version of CenterForce Analyzer also lets you and additional users set schedules for e-mailing scorecards to individual agents and groups of agents. Previous versions enabled only one user to set schedules for sending scorecards on a regular basis, such as daily, weekly or monthly.
CSG Systems' CSG ProfitNow! generates forecasts to predict customer behavior based on their purchase histories and conversations with agents. Agents can learn which customers might abandon your company. And they can view suggestions on how to win back their trust. The software also identifies your most profitable customers and recommends what additional products and services agents can offer them.
Agents view color-coded signals from their PC desktops to see the overall level of customer satisfaction. A red alert signals a high likelihood of customer abandonment. (Red alerts are based on data like how often customers call to complain and a history of difficulties they've had with your company's products and services.) The yellow signal indicates a reduced risk of customer attrition; green and blue denote the more profitable customers. Agents view strategies for handling each type of customer based on rules that you set. They might offer apologies and discounts to red-alert customers or pitch new or more expensive products to customers within the green to blue spectrum.
CSG ProfitNow! collects data about customers from sources like billing software and CRM software to create forecasts about customers based on their purchase habits, how frequently they call and what topics they discuss with agents, among other factors. If the software finds that some customers are dissatisfied, it predicts how they might distance themselves from your company for up to 28 days. The software projects what products and services happy customers might request next for up to 56 days.
You can use CSG ProfitNow! as a standalone product or as a module of CSG Advanced Customer Service Representative (ACSR), a CRM software suite. If you license CSG ProfitNow! without ACSR, the software comes with CSG Configurator. CSG Configurator lets you integrate CSG ProfitNow! with billing and CRM software to collect the data needed for forecasts. CSG Configurator also enables you to enter the suggestions that agents view when they help the most disgruntled and most valuable customers.
Informatica lets you build an analytics suite one module at a time. You can begin with PowerCenter 6 to collect historical analytics and PowerCenterRT for real-time analytics. Both versions of PowerCenter work with Business Adapters series of modules that enable PowerCenter to collect data from sources like phone switches and CRM software from vendors including Oracle, PeopleSoft, SAP and Siebel.
You can use Informatica Warehouse to save data from both versions of PowerCenter. Informatica Warehouse automatically organizes that data into tables. The software lets you view separate reports that contain information like sales figures, how customers respond to marketing campaigns, and the content and frequency of customer support calls and Web-based communications. Informatica Warehouse also includes 50 reports for generating metrics based on the data that you save from PowerCenter.
Informatica's PowerAnalyzer lets users view data from a Web browser-based dashboard. The software includes a setup wizard to define what analytics to display and how often to display them. To learn how many customers from a given location are responding to a promotional offer, you can set rules using the wizard to view hourly updates on call volumes for customers from that venue.
The analytics options available with knowledge management software can provide you with valuable insights about customers based on their on-line behavior. OneStep from iPhrase (Cambridge, MA) lets you find information that customers request with User Summary, User Intelligence and Content and Knowledge Gap reports.
User Summary reports enable you to determine the success of customers' on-line searches based on the answers they receive. These reports indicate how often OneStep provides customers with text answers; routes customers to relevant Web pages; and generates dynamic reports containing information from Web pages and knowledge base content in response to a search request. User Intelligence reports the most popular customer search requests and the answers that OneStep provides most frequently to customers.
Content and Knowledge Gap reports let you determine which search requests OneStep doesn't provide answers for. You can identify the words that customers enter for each unsuccessful search. And you can enter those words into OneStep so the software can recognize them to complete future searches successfully.
Content and Knowledge Gap reports also include Buried Content, a report that displays how many pages of your Web site customers view, starting from your home page, to find information they want. Buried Content provides an analysis of Web site visitors' behavior that can help you re-design your site more efficiently. So if you find that most visitors click to the third page of your site to download a brochure, you can decide to include the brochure on your home page for more immediate access.
Another vendor of self-service software, Island Data, offers Insight RT. The product's Concept Recognition Engine analyzes the text of customers' e-mail messages and on-line forms that they fill out. Insight RT recommends an action based on its analysis of the text.
You enable Insight RT to assess the content of customers' messages by entering "seed words" from a setup wizard. For example, it's not often that customers write, "I want to buy your product." They're more likely to write something like, "How much does this product cost?" By entering the product name and "cost" as seed words, Insight RT can recognize them in the body of an e-mail message or an on-line form and identify it as a sales lead.
The software also includes a lexicon of words and analyzes the statistical likelihood that specific words will appear together in the same message. When you enter seed words, the software presents you with sample sentences in which those words might appear. If you enter "subscribe" and "cable TV service," the software will automatically identify sentences like "I want to subscribe to your cable TV service." You select which sentences accurately represent what customers are likely to say.
Insight RT then analyzes text from customer feedback and rates each sentence with a percentage value to indicate the likelihood that it's a specific type of message. A sentence as clear as "I want to subscribe to your cable TV service" could receive a value of 100 to categorize the message as a definite sales lead. You can then set rules to enable agents to take specific actions based on these scores. If you integrate Insight RT with CRM software, you can evaluate a large volume of customer inquiries, and only insert the highest-scored opportunities into the CRM queue for agents' immediate attention.
The Dashboard feature of Insight RT lets you track data like customers' on-line browsing behavior and purchasing habits. Dashboard updates this data every ten to 15 minutes and displays it from customizable graphs and charts that call center managers and your company's sales and marketing departments can view from their desktops. The software can also generate reports containing historical data; and can integrate with third-party reporting software like Seagate Software's Crystal Reports.
Jeeves Solutions (Emeryville, CA) lets you determine how effective your Web-based self-service is with the Analytics module of the company's knowledge management software, JeevesOne. Analytics include templates for 107 reports that can encompass the number of self-service sessions that customers successfully complete, abandon or escalate to receive assistance from agents.
Your reports can also contain the content of customers' on-line searches and the information that JeevesOne provides them after each search. You can also add customer profiles to each report to determine which visitors to your Web site who use the self-service option are new or repeat customers; are from specific geographical regions; or have purchased selected products and services.
Analytics recognizes specific words and sentences, or paraphrases of them, which customers enter to search an on-line knowledge base that you create using JeevesOne. Analytics then assigns percentage values for each customer search. You set thresholds based on these values so that if Analytics assigns a given percentage value to customers' search words, agents receive e-mails containing a list of the words and phrases customers use for their on-line searches. If you set a threshold of 90% for e-mails that contain possible sales leads, the e-mail agents receive could include customers' questions about the pricing and warranty for a given product.
You can improve the accuracy of each customer search from JeevesOne by entering into a setup wizard the words, phrases and sentences that customers are most likely to use in a search. The software generates paraphrases so that it can recognize different ways of requesting the same information. If you find that customers are phrasing their requests for information in a way that JeevesOne didn't predict, Analytics lets you view the verbatim inquiries that JeevesOne didn't recognize and prioritizes them based on how frequently they occur.
For monitoring and improving agent performance, Emvolve from Performix Technologies includes the Performance Review module. Performance Review provides agents with a scorecard they view through a Web browser to see how closely they meet call center goals, like sales quotas or the number of customer service requests they resolve daily.
Emvolve's Key Performance Indicators offer you a global perspective of your call center's performance statistics. You click on a particular statistic to view additional details. If you notice a rise in the number of call abandonment rates at your center, you can click on that figure to find the abandonment rates for each team of agents. You can then click again to find the same information about individual agents working for a given team. The software also works with multi-site call centers so that you can measure the performance of call centers from separate locations.
Other modules in Emvolve help you to rate agents' performance. Competency Review allows you to assign scores for each agent based on categories that you set, like politeness and attentiveness. You can also allow agents and their co-workers to score themselves and each other to gain insights about what they think their strengths and weaknesses are.
To further sharpen agents' performance, Personal Development automatically indicates the areas that agents need to improve upon, like sales technique and quality of service. And if you want to encourage agents to continue their good work, Reward and Recognition calculates the bonuses that agents receive for meeting the goals that you set with Performance Review.
The newest version of Quadstone's Quadstone System suite works with J2EE and Java 3D so that you can set the analytics you want to keep track of from a wizard and view them in 3D charts. Quadstone System also supports XML so that your company's IT staff can create customized interfaces, enabling the software to collect statistics from multiple sources.
Decisionhouse, the centerpiece of Quadstone System, lets you view information about customer behavior. You can, for example, enter an inquiry as basic as tracking the number of calls agents answered for a given day, week or month; or as specific as the number of customers from New Jersey who asked technical support questions on Mondays from 9 am to noon. Decisionhouse then displays the data you request as a customized chart or graph.
The Transactionhouse module collects the data that you view using Decisionhouse from sources like phone switches, databases containing customer information and CRM software from vendors like Onyx and Siebel. Transactionhouse can also track the Web pages that customers visit.
Rounding out the Quadstone System, Actionhouse lets agents automatically take specific actions based on thresholds you set and on data that Decisionhouse collects. The software also integrates with other vendors' software to enable agents to take these actions. You can, for example, set a threshold for a maximum hold time of two minutes using Actionhouse. But if Decisionhouse finds that some customers remain on hold for more than two minutes, Actionhouse can work with your CRM software to provide agents with a call script, including an apology for the delay in response.
SAS Interaction Management lets you track individual customers' behavior during a timeframe that you set, like how often they call your company on a monthly basis. The software can additionally collect data about customers, like histories of their purchases and their communications with agents.
SAS Interaction Management enables agents to take actions based on predefined rules by selecting options from a drop-down menu. Using the previous example of tracking customers' monthly calls, you can set a rule requiring that agents contact customers who've made a marked rise or decline in their monthly call volumes to inquire if they're satisfied with the service they've received.
You can use SAS Interaction Management with SAS Marketing Automation, software to identify target audiences for outbound marketing campaigns. The software provides you with a Java-based user interface to determine which customers might respond to a given offer, based on their prior purchases. SAS Marketing Automation suggests the best times and dates to contact customers. And it recommends what products and services might appeal to them.
Unica's Affinium Model comprises four modules featuring sales-related analytics. The tools help you determine who your most valuable customers are and how to make interactions with each more profitable. The software is available as a standalone product or as a module of the Affinium software suite for predicting customer behavior and designing marketing campaigns.
The Customer Valuator module enables you to predict the lifetime value of customers based on their demographic profiles and previous purchases. Response Modeler applies the same data to help you answer questions about customers (e.g., whether they pose a credit risk; are likely to discontinue their business relationship with your company; or whether they'll take advantage of a promotional offer).
You can define customer demographics with Market Segmenter/ Profiler based on the specific products or types of products that customers purchased and how much money they spend. Rounding out Affinium Model's analytics options, Cross Seller lists customers' purchases and predicts what items they'll probably consider buying next.
We want to hear from you! Please e-mail: lhollman@cmp.com.
Self-Service Analytics
Knowledge management software lets visitors to your Web site find information on their own. When combined with analytics software, knowledge management applications can improve visitors' self-service experiences to reduce your company's overhead and increase customer satisfaction.
"Analytics allows you to take the guesswork out of determining what customers are trying to find by providing a way for you to examine exactly what information your customer is attempting to access," explains Andre Pino, senior vice president of marketing for iPhrase (Cambridge, MA).
For example, consider how pharmaceutical firm Novartis used Jeeves Solutions' (Emeryville, CA) JeevesOne software with the Analytics module. Novartis observed high abandonment rates among knowledge base users and a subsequent rise in call volumes. The company reviewed the text of customers' unsuccessful searches that users conducted before calling agents about the side effects of prescription drugs. Novartis then entered the words and phrases that customers used for those searches, enabling JeevesOne to recognize them so that customers could more easily find information about side effects.
James Speer, director of product marketing for Jeeves Solutions, says that Novartis provides a textbook example of how you can use analytics software to improve Web-based self-service.
"The big thing is to look at the number of interactions where the overall session was less than desirable," he explains. "By tracking the [self-service] sessions, if you find that 40% of transactions result in escalation [to agents], those are the key metrics you need to work with."
But no knowledge base can contain all the answers. "The customer should be allowed to go to the next best possible options for service, which include an escalation to an agent," says Norm Williams, president and CEO of Island Data (Carlsbad, CA).
Need More Data About Analytics Software?
Take a closer look at the analytics products mentioned in this feature by contacting the vendors below.
CenterForce Technologies
CSG Systems
Informatica
iPhrase
Island Data
Jeeves Solutions
Performix Technologies
Quadstone
SAS
Unica
|