Improving customer interactions: Finding the root cause of customer dissatisfaction with speech analytics

White paper
July 2009


Executives responsible for customer relationships often have difficulties understanding why customer retention rates are declining at their organization. After reviewing sample call recordings to assess agent effectiveness, they may decide to retrain selected staff members to improve their telephone skills. Despite the extra training the numbers may continue to drop because they're not looking deep enough. If they had a way to analyze all customer conversations, they would have discovered that the root problem wasn't the quality of the agents, but a complicated process that annoyed customers enough to convince them to try the competition.

Companies are under increasing pressure to deliver the highest quality customer experience. Competition for customer mindshare is fierce, while at the same time loyalty is more elusive than ever. In many cases, one poor customer experience will have them switching to the competition without a second thought.

Industries with high customer turnover understand that consistent quality customer experience at every point of contact is a critical differentiating factor that cannot be taken for granted. Quality monitoring of customer interactions is therefore key to competitive survival.

It is essential that operations have the means to detect potentially unsatisfactory situations and take the steps to correct them. Once the corrective action is taken, it is also important to measure the effectiveness of that action.

Quality monitoring in a customer service contact centre is nothing new. The vast majority of companies that operate call centres have been using systems for recording voice conversations for some time. While this provides some degree of quality control, it has a number of limitations.

This approach only provides a partial picture of what is happening for two reasons. First, it entails reviewing only random samplings of conversations (on average less than one percent of all conversations). Second, the focus of this approach is on monitoring the quality of the customer service representative's work, which in a majority of cases, is not the root cause of the dissatisfaction.

Statistics from TARP, a US-based customer service research firm, show that the root causes of customer dissatisfaction with their contact centre experience are broken down as follows:

  • 20 percent to 25 percent are based on the quality of the representative's work – this includes failing to follow procedures, knowledge gaps and behavioural issues
  • 55 percent to 65 percent are a result of the company's processes – this can include product/service deficiencies, marketing policies, billing errors and broken processes
  • 15 percent to 20 percent are a result of customers' mistakes or unrealistic expectations

It is clear from these numbers that enterprises must look beyond the narrow confines of the customer service representative's performance to discover the “why factor” in customer satisfaction. That means looking at the end-to-end customer experience and business performance in their quality monitoring practices.

Organizations that can move through this continuum are able to:

  • Isolate the root causes of customer dissatisfaction
  • Address problem areas proactively
  • Identify gaps in knowledge or business processes
  • Improve voice and Web self-service
  • Improve the quality of service delivery
  • Improve the effectiveness of customer feedback
  • Improve up-selling and cross-selling opportunities
  • Obtain quick customer feedback when launching new marketing campaigns
  • Retain more customers and reduce churn
  • React more quickly to competitive initiatives

Taking quality monitoring to the next level

Given that the majority of customer interactions take place via voice transactions, the information that can be extracted from conversations is absolutely invaluable in understanding the customer service experience. At the same time, to realize process improvements and/or pinpoint new business opportunities, organizations must engage in strategic process improvements that go beyond random samplings of recorded conversations.

This is no small feat when one considers that a conversation is a huge, unstructured mass of information that does not easily lend itself to categorization. The challenge therefore lies in the ability to capture and analyze that wealth of information efficiently and effectively to deliver meaningful results. This ability can be found in speech analytics.

Speech analytics is a combination of tools and processes designed to measure the quality of all aspects of service delivery revealed in conversations. It allows organizations to organize and categorize call content to highlight situations that lead to customer dissatisfaction.

Speech analytics technology is used to mine customer conversations for hidden insights into customer needs in order to determine root-cause issues of dissatisfaction. It captures customers' conversations and transcribes/transforms them into key elements for categorization and prioritization. This enables organizations to target problem areas based on all conversations and drive actionable results.

Data capture and categorization however is only one step in speech analytics. In order to realize the full value and enable strategic process improvement, it is essential to leverage that information as part of an end-to-end process. This process includes everything from data collection through to planning and execution of solutions to ongoing monitoring and measurement.

Speech Analytics from Bell

The Speech Analytics consulting service from Bell is part of a series of engagements aimed at improving the quality of our clients' interactions with their customers and the overall customer experience. So far, we have mainly discussed customer dissatisfaction and causes linked to defective company processes. There are other circumstances in which Speech Analytics can help improve customer relations. Here are a few:

  1. Understanding WHY customers call
    Identify the reasons why customers call and the processes in place that drive those customer interactions. Identify the process frustration, the accountable group, determine the corrective action, and measure its effectiveness.
  2. Improve voice self-serve
    Identify call categories that could be automated and for which the volume is sufficiently high. Monitor calls to better understand how customers express themselves when looking for services.
  3. Improve Web self-service
    Identify calls that are driven into the contact centre because the Web self-service did not meet customer expectations. It could be a situation where the customer did not find the information he was looking for or was unable to complete a transaction. Monitor these calls to better understand customer behaviour and impact on overall loyalty.
  4. Improve the agent work quality more efficiently
    Identify the calls associated with customer frustration due to insufficient training given to the agent. Determine the root cause of the problem for improvement in areas like agent scripting, new processes, etc. Also identify high-quality calls to use as a contrasting example and leverage in new training situations.
  5. Improve the immediacy of customer feedback
    Many companies hire specialized firms to survey customers. By having access to the entire conversation, they can understand the reason behind customer dissatisfaction and frustration on an ongoing basis. Think of this as daily customer focus groups.
  6. React quickly to competitors' initiatives
    While speaking to representatives of your organization, customers may provide valuable information about competitors and the services or products they offer. Analyzing conversations that discuss competition may help companies react quickly to shifting market offers.
  7. Improve the effectiveness of marketing initiatives
    Some companies regularly launch marketing initiatives that encourage customers to participate either by voice or Web self-serve or with the help of an agent. In most cases, customers who are confused or cannot complete a transaction will call an agent. These conversations can be isolated and used to explain the success or failure of a marketing initiative.
  8. Improve the effectiveness of cross-selling initiatives
    Companies are increasingly encouraging their agents to use cross-selling techniques. Selective call monitoring helps them understand why the sale succeeded or failed. Based on the findings, the products or agent solicitation scripts could be quickly improved.

As a leader in contact centre services and speech analytics development and deployment, Bell has established a proven methodology to help organizations derive maximum benefits from conversation analysis. This process includes the following steps:

  1. Discovery session
    The discovery session begins the process. Its purpose is to help the client better understand the key drivers behind selected business issues. Part of the process involves information gathering on existing metrics and known customer frustration points, as well as identifying key resources from the client's organization. Given that customer dissatisfaction may arise from a company's processes, participants must be selected based on their ability to take corrective action and assume responsibility for it.
  2. Data extraction
    This step involves extracting customer interactions from the client's existing recording system (A test extraction is done to ensure compatibility and recording quality).
  3. Identification of categories
    Once the calls are extracted and converted into text, our consulting experts spend time building call categories and key word hypotheses that are then mapped against the calls for accuracy and frequency.
  4. Focused Call Listening and Analysis
    Calls are then sorted into categories – call drivers. Our consultants then listen to the specific points of the customer interaction to determine accuracy and identify secondary call drivers that lead to customer frustration. Analysis of the calls is ongoing through this phase and provides the basis for our report presentation that identifies the call drivers affecting the business issues.
  5. Presentation of the Analytics Findings
    Experts choose conversation excerpts that faithfully reflect a particular situation that generates dissatisfaction; these excerpts are the “voice of the customer”. Bell specialists will suggest potential causes for the situation, and the designated group will develop an action plan.
  6. Ongoing Analytics Program
    After corrective actions have been taken, it is important to continue with further selective listening, to validate whether the action has been effective.

How Bell can help – a working example

An organization focused on improving quality and the overall customer experience for its services found that incoming customer calls were generally a result of a broken business process or poor customer communication in other departments. Bell entered into a speech analytics consulting engagement to find ways to determine the root cause of the quality of service issues and to ensure an ROI on the enterprise's investment in speech analytics technology.

The goals of the engagement were to:

  • Identify challenges created by the enterprise that drove call volume and their impact on customer dissatisfaction
  • Expand quality monitoring to identify “early warning” trends
  • Expand existing quality initiatives to the enterprise level
  • Identify the top causes of customer dissatisfaction
  • Improve operational efficiency and customer experience

The process included the following steps:

  1. Engaging executive sponsors in identifying goals, educational needs and quality initiatives
  2. Identifying and briefing key stakeholders across the organization (e.g., Marketing, Communications, Training, Contact Centre, New Product Development, etc.)
  3. Using analytics tool to develop call categories, keywords, strings, events, call exceptions, trends, etc. to form the foundation for her in-depth analysis
  4. Conducting quarterly sessions to review top three quality issues, supported by multiple clips from customer recordings
  5. Providing action plans to improve and work within each of the departments identified on the top customer dissatisfiers for that quarter

The engagement led to the following improvements:

  • Increased customer loyalty scores
  • The establishment of multiple process improvement initiatives
  • A renewed value in agent scripting and adherence
  • Shared ownership of the customer experience beyond the contact centre
  • Confirmation that ROI was attainable within 12 months
  • Greater departmental cooperation with the contact centre
  • Improved customer communications
  • Took the voice of the customer to the executive level

About Bell ICT Solutions

Bell is a leader in driving ICT convergence strategies for enterprises. With a world-class, proven ICT infrastructure -backed by a cross-functional team of industry-leading experts and more than 125 years of business success -we are unique in our ability to provide flexible, end-to-end solutions and services that can be customized to meet evolving business demands.
We offer:

  • The power, performance and reliability of the country's largest IP infrastructure
  • State-of-the-art network, data centres, lab and technology resources
  • In-depth knowledge of business process solutions and applications for different industries
  • Consistent levels of service, performance and solution reliability
  • A strong commitment and track record of performance in developing customer-focused solutions

For more information visit, or request to be contacted by a Bell representative at