How Call Center Speech Analytics Lower Costs and Increase Customer Satisfaction


  • Call center speech analytics harvest call recordings to provide customer and agent insights not available from any other source.
  • The technology delivers sustained benefits in 6 important areas, all of which contribute to improved customer satisfaction and substantial cost reductions.
  • Experience shows that speech analytics deployments achieve an average payback period of 11 months and an average ROI of 26 percent beginning in the second year of operation.
Johnny Grow Revenue Growth Consulting

The telephone is still the top channel for customer support. The volume of phone calls creates a treasure trove of customer data that can yield customer insights and contact center performance improvements if unearthed. An increasing number of customer service leaders are using speech analytics to do just that.

Speech analytics is a software technology that analyzes real-time speech or voice recordings to gather customer and case insights.

The software converts unstructured audio conversations into structured data that can be used for call or contact center process improvements and information reporting.

Contact center speech analytics detect emotions expressed by both customer and agent and measure case resolution performance factors such as hold times, verbal escalations, dead air periods or agents talking over customers.

This customer service technology detects keywords, phrases, patterns, pitches and periods of silence to measure customer sentiment and the outcomes of support calls. The data allows contact center managers to identify, categorize and tabulate what's working and not working to achieve their goals of customer satisfaction and operational efficiency.

It can also be used for speech recognition to work with call routing technologies such as IVRs, and improve IVR performance by reducing or simplifying complex touchtone menus. Speech recognition can also be used with intelligent chatbots to enable customers to use voice commands to lookup information, get an answer, perform a transaction or get routed to the right destination.

Contact Center Speech Analytics

An Accelerating Trend

Customer service technologies are widely adopted when they deliver results. Analyst and market research show that contact center speech analytics became a billion-dollar industry in 2020 and continues to grow at double digits. These solutions use automation to detect real-time case variances, suggest near real-time agent performance improvements, monitor compliance requirements, extract customer insights and lower call center costs.

Consumer-based industries such as telecommunications, financial services, retail, and travel and hospitality are top adopters. But B2B industries are increasing their use as the benefits are no less applicable to them.

An Opus research survey of support organizations using contact center speech analytics found that 72 percent said these solutions contributed to improved customer experience, 68 percent said it reduced call center costs and 52 percent said the technology leads to revenue improvements.

Research reveals that call center speech analytics contributed to improved customer experiences, reduced call center costs and revenue growth.

Click to Tweet

Benefits, Payback and ROI

We have all heard the recording that states "your call may be recorded for training and quality purposes." Well, it turns out there are far more uses and benefits for these audio recordings. Based on our experience of implementing speech technologies and measuring the payback, we group the benefits into six categories.

Speech Analytics Benefits

Customer Insights

  • The technology measures why customers call and what they are looking for. This aids preparation for timely case responses and can forecast agent staffing requirements.
  • It also identifies the steps, processes, agents and other factors that correlate to successful and unsuccessful customer outcomes. This is the data that shows how to improve processes to achieve more predictable customer outcomes.
  • It can detect vocal tone, fluctuating amplitude and certain customer emotions to identify customers at risk of churn, so that agents can shift their approach or escalate to customer retention specialists.
  • It can sift through calls for keywords such as competitor names or threats to cancel, changes in customer sentiment or repeated themes over multiple calls to select which calls warrant management attention.
  • Sentiment analysis determines whether customers are happy, angry or indifferent and thereby can trigger relevant and personalized resolutions, offers or escalations.

Agent Insights

  • Real-time speech interrogation can provide agents with suggestions, prescriptive guidance, potential solutions or recommended next-best-actions. The technology can advise what question to ask next, what solution to try or remind agents to read a legal disclaimer.
  • These solutions can identify areas where agents are the least comfortable or may have the least knowledge and where training can deliver quick improvements.
  • When agents are performing poorly their performance is normally not recognized until enough dissatisfied customers have come forward. As most dissatisfied customers will remain silent and simply lessen their business or defect, the negative impact to the company is exacerbated over an unrecognized and extended period. These solutions detect and report these situations in real-time so course corrections can be applied before they generate sustained financial losses. Or on the flip side, the analytics also detect successful efforts in real-time to recognize agent excellence.

Case Insights

  • The technology can rank what agents, actions or processes lead to the most and least successful case outcomes. This can optimize call routing and process optimization.
  • Speech software can compare steps, scripts or knowledge artifacts that generated the highest and the lowest customer satisfaction (CSAT) scores. This information can be used to eliminate process inefficiencies, refine procedures, suggest training or identify coaching opportunities.
  • The technology can measure case types, quantities and trends. This helps plan training and agent scheduling.
  • It can measure case response method effectiveness. For example what scripts, knowledgebase articles or other artifacts are most helpful in resolving each type of case.
  • These solutions can identify repeated cases, similar cases or case patterns to show where root cause analysis and resolution will deliver the biggest payback.

Quality Assurance

  • Speech technology can identify and escalate customer conversations that require management attention.
  • The software can detect keywords that suggest data privacy, information security, regulatory or compliance risks, and should be reviewed.
  • Without this technology, assessing agent calls to pinpoint regulatory or compliance problems takes so much time that only a sampling of calls is really done. This creates a significant risk. Applying technology automation is often the only way this necessary process can occur with confidence.

Customer Service Analytics

  • It's not realistic to think that agents are going to share every call detail for customers that cancelled, delivered low customer satisfaction (CSAT) scores or incurred similar unsuccessful calls. Nor are they going to share fully objective information as to why average handling time (AHT) is long or speed of answer (SoA) is slow. These answers should be derived from data that considers many calls, not a select few. When the data is accumulated, and filtered by agent, call type, customer type or other variables, managers get the information they need to make timely course corrections. Without this data, managers are running blind.
  • Customer service analytics identify the calls with long AHT by analyzing excessive silence or multiple wrong attempts before finding a solution. They identify repeat calls and examine repetitive incorrect resolutions before getting to the right answer. They identify what scripts worked and which did not. In these examples, they deliver the data to lower AHT and increase First Call Resolution (FCR). They can similarly detect the types of calls that are completed very quickly in order to suggest which calls may be pushed to customer self-service channels.

Cost Savings

  • Using speech recognition for call monitoring and compliance verification saves hundreds of hours quarterly and reduces headcount.
  • When the technology identifies the contributing factors to long AHT or low FCR, actions are taken to reduce cost per contact and increase CSAT.
  • Process optimization, such as when to use scripts or what scripts work the best, improve agent productivity and lower labor cost.
  • Agent specific recommendations for education, training or coaching will improve agent productivity and tenure, and lower recruiting and hiring cost.
  • Best of call, the cost savings are not at the expense of customer satisfaction. In fact, quite the opposite. The improvements from speech technologies lower cost to serve while increasing case outcomes and customer satisfaction.
  • Our experience in implementing speech technologies with clients is they achieve an average payback period of 11 months and an average ROI of 26 percent beginning in the second year of operation.

Speech Technology Challenges

As with most customer service technologies, the benefits are not without challenges. So, when evaluating speech analytics, it is important to know the risks and mitigating factors.

  • Selecting the best solution can be a confusing process. To prevent becoming confused on vendor hype, alleged benefits and technology claims, and to enable an apples to apples comparison among the vendor solutions, it’s best to prioritize your pain points, measurable objectives and system requirements before you begin talking to vendors.
  • The software is not cheap and the implementation is not fast. At least not if you are going to achieve sustained benefits and payback. Software configuration requires technical expertise and most organizations will require significant system integration. Once the implementation goes live, it will require at least a few iterative refinements to achieve its potential.
  • I'm going to double down on the system integration challenge. Most contact centers already have way too many fragmented and disparate support applications. They don't need yet another. System integration will be needed to sync audio with the CRM contact, activity and case records and contribute to a 360-degree customer view. Additional integrations may be needed for customer portals, mobile support or other support channels. These integrations are essential to maintain or improve agent productivity. Several speech technologies offer packaged integration with popular CRM systems such as Microsoft Dynamics 365 and Salesforce. However, some are known to be shallow or incomplete so due diligence is needed.
  • Also related to implementation, many customer support organizations underestimate the number of resources needed for deployment. That mistake will contribute to implementation time and cost overruns. Ensure a comprehensive project and staff plan before you commit.
  • It can be difficult to build a business case that supports the technology investment with an accurate ROI. Industry recognized benchmarks are needed to quantify the hard numbers that drive an achievable financial payback.
  • Some managers struggle to make the analytics actionable. Without action, information reporting is merely interesting. To induce action, we typically limit the number of key performance indicators (KPIs) to only the most influential and we display them in dashboards with benchmarks and predictive analytics that bring visualization to the numbers. If after these measures the data is still not acted upon timely or at all, managers may require training, or a broader change management program may be needed.

For most customer support organizations, recorded conversations sit idle and untapped. That's unfortunate as this technology can use that data to provide customer and call center insights not available from any other source.

Speech Analytics Call Center