There is a growing amount of technology available in both the call quality monitoring (QM) and customer analytics arenas. A lot of it is expensive and often seems to be a solution looking for a problem. On the other hand, call center quality is still lacking in a majority of contact centers and most executives would agree that at least 30% of their workload was preventable if the rest of the company paid more attention to the voice of the customer. These two situations exist because most executives have not fully understood how to use monitoring and analytics to address both quality improvement and workload prevention.
Before we examine the newly available technology, let’s review why you are monitoring and gathering voice of the customer (VOC) data.
Compliance Is the LEAST Important Reason for Monitoring
When contact center quality and productivity are not where management would like them to be, a standard response is, “Let’s fix the monitoring sheet and the monitoring tools.”
There are five pitfalls in this approach unless it is executed very thoughtfully. First, there may not be a complete understanding of the drivers of satisfaction for major types of calls being handled. Secondly, supervisors may not be able to effectively use the results. Thirdly, the incentives for the managers and outsourcers may not encourage adoption of the processes to use the output. Fourth, the response rules and tools available to the staff may not be able to successfully satisfy the callers leading to costly escalations and dissatisfaction. Finally, the company’s sales messages, communications and policies may not be leading to correct caller expectations and an ability to efficiently create a satisfied caller. Therefore, you must decide how broadly you want to address the context within which the QM tools will be used.
An effective QM process should have six impacts:
- Provision of immediate feedback to a CSR.
- Guidance to supervisors on where individual CSRs need additional coaching, as well as instances meriting positive recognition.
- Input into team and center training and call-flow improvements.
- Intervention with customer after the fact on calls receiving low scores where serious damage has been done or misinformation given.
- Identification of response rules that should be enhanced or modified.
- Guidance on needed client improvements in operations and policies outside the call center.
The last two of these impacts are seldom viewed as necessary outputs of a QM system, but often present the biggest opportunity for significant improvement. The QM data is a critical part of the overall voice of the customer, which is the key input to improvement of the overall, end-to-end customer experience. However, it must be integrated with other data if it is to be actionable and allow rational priority setting. For instance, in most call centers, CSRs with even the most basic training can successfully handle 50% to 60% of calls. Where coaching and feedback is necessary are in the more difficult and nuanced calls. Therefore, focus on this subset of calls is required. Further, the subset is made up of dozens of types of calls, each of which has its own key factors. The QM data must be integrated with automatic call distributor (ACD), satisfaction surveys, and customer relationship management (CRM) system data to allow linkage to wait time, resulting satisfaction, and reason-for-contact data, as well as operational data describing the customers’ transaction history with the company.
What You Should Expect from QM Technology
The major new technology in the QM arena is the ability to digitally record all calls and emails, label it with assigned reasons for call and product information and then analyze the conversation via text or speech analytics to understand the content. The more advanced tools allow you to not only do searches on key words or phrases, but also to understand the intent of sentences and the outcome of calls in terms of resulting satisfaction. This same analysis can be done manually, but the technology allows you to process many calls faster. Further, the technology can flag calls real time that are going off track or resulting in anger. This allows for intervention.
Case in point: A leading appliance manufacturer identified that repair costs were rising for certain appliances, including multiple visits to consumers’ homes, leading to decreased customer satisfaction. By combining data from call monitoring, speech analytics, reason for call, product symptom and the operation service visit record, the VOC team was able to identify that the repair technician often arrived with a fuzzy idea of the problem and then found that he lacked the correct part to complete the repair. Analysis of the call record and the actual call recordings found that the CSR was not taking enough time to obtain and record the detailed symptoms of the problem, which would have assured that the technician had the necessary parts on the first visit. When this additional information was gathered during the call (which often took less than a minute extra), the percentage of service calls requiring a second $80 to $100 truck roll was reduced by more than 20% and customers’ satisfaction for that category of calls rose double digits.
VOC Analytics Is Much More (and Much Less) than Frequencies and Cross-Tabs
VOC analytics should not only report volumes, but also give you an insight into why things are happening and which events merit your attention. Analytics should take the 20 data points on 50,000 calls (a million data elements) and tell you the three things you should be worrying about. To understand the implications of these two approaches, picture two reports:
- The first is a report containing five bar charts and five cross-tabulations describing the content and satisfaction of 10 types of calls about five products (a 10-by-5 matrix).
- The second report is a one-page memo describing the three most important issues the company faces, the volume of customers encountering each of the three issues, the revenue damage and cost for each month action is not taken, and a recommended course of action for each issue.
Which report will get the most action the fastest? Further, what if each of the three issues was highlighted in three separate individual reports sent to the strategic business unit head who was responsible at the same time the combined report was sent to the CEO? Tailored reports for each function always get more action than combined reports and masses of data tables. Your analytics tools should help you combine everything you know about an issue, and then discard everything that doesn’t apply to the important issues.
Analytics should include four specific types of combined VOC and QM analysis:
- Real-time analysis to identify calls requiring immediate intervention;
- Basic analysis to support basic reporting based on linkage of QM to surveys.
- Outcome: first-call resolution, caller satisfaction, preventive education—based on observation and or survey.
- Performance: security verification, accuracy, quality, courtesy, professionalism, rapport, education.
- Efficiency: talk-time ranges for broad categories of calls, control of conversation.
- Opportunity analysis to identify systematic improvements. This should identify issues with:
- systematic low satisfaction or inefficiency, implying defective response rules, repeated escalations;
- requirements for call-backs due to lack of information; and,
- repeat calls from the same customer about the same claim or type of claim.
- Progress on ongoing identified issues. This would identify what issues are actually resolved, where the needle moves, or ideally, whether the issue disappears.
Steps for Moving Toward World-Class Service
The following are the seven steps for using QM and VOC technology to move your call center to world-class service.
- Confirm drivers of satisfaction. An overall set of average drivers may be misleading. For instance, in auto insurance, 75% of calls are from customers without a claim; mainly billing and account maintenance. The other 25% of calls are from customers with a recent claim. In the latter case, the drivers of satisfaction and what should be measured via monitoring are very different.
- Determine the sample to be measured. Most organizations draw a random sample of calls. TARP has found this can be a huge waste of resources, as most short calls are simple and not a cause of either dissatisfaction or waste. On the other hand, if all calls are recorded and there is no per-call cost to analyzing all calls using a mass analysis tool, then 100% of calls can be analyzed.
- Collect the data via manual and or automated methods. This activity includes the actual recording and labeling of the contacts so that they can be analyzed. The more extensive the labeling is, the more actionable the analysis will be to allow achievement of all the levels of impact described above. However, in most organizations the labeling can create significant systems integration demands.
- Collect data using the best QM data collection method. Critical issues in selecting the appropriate method include where the recorded calls will be stored and the extent of labeling possible. The collection tool can stand separate or be integrated within the analysis tool.
- Conduct an integrated analysis producing the four types of analysis noted above. Those factors within the control of the call center (individual CSR adherence and needs for enhanced general training and call flows) should be separated from those external to the individual call center (changes in response rules and changes in policies and procedures in the rest of the client organization). Even this most sophisticated automated analysis still requires human interpretation and translation into bottomline implications and recommendations.
- Conduct multilevel reporting. Reporting must be tailored to each level of management and function if it is to have impact. Reporting is ideally exception based and actionable in that it not only highlights problems and stellar performance but it also suggests plans of action to address them. Further, it creates an incentive for action by quantifying the cost of inaction either in dollars or declining customer satisfaction.
- Suggest action plans and targets for call centers and overall organization. Suggesting a plan of action as part of the voice of the customer significantly reduces resolution cycle time because the recipient has a place to start. Technology, to be cost effective, must be viewed as a tool for implementing a well-understood methodology, not as an end in and of itself.
John Goodman is Vice Chairman of TARP Worldwide, and author of Strategic Customer Service.
– Reprinted with permission from Contact Center Pipeline, www.contactcenterpipeline.com