Put simply, speech analytics is the process of analyzing the voice of the customer by collecting the wealth of retrievable, identifiable and reusable information from every interaction between that customer and a company.
Companies traditionally use speech analytics in customer service to deepen their understanding of the voice of the customer, and it has become an essential lever for customer relationship management, especially in call centers.
Speech analytics can go further when combined with quality management or quality monitoring, i.e. analyzing the agent's voice as well as the customer's, which supports both quality control and compliance checks on every customer–agent interaction.
The field of emotion & speech analytics now adds an extra dimension to speech analytics: human emotion. Compared with a traditional system that combines written transcription with keyword search, the emotional factor gives customer service managers a finer understanding of call reasons and customer pain points.
While speech analytics analyzes the words you say, emotion & speech analytics technologies analyze the non-verbal content beyond speech (prosody): how customers and agents interact, including the physical cues that signal emotions such as strong anger or delight. These are then linked to topics and call reasons so companies can prioritize improvements to their processes, products and services, or to agents' arguments and scripts.
Planning a speech analytics project?
See what Batvoice reveals about your own calls, with a quality expert.
In time spent by agents and supervisors on low-value calls, digitalized first.
Less time spent on your quality processes: calibration, call selection, evaluation and debriefs.
Time needed for each quality monitoring evaluation, freeing up just as much time to analyze results and help agents with their training.
CRCs (customer relationship centers, known in French as "centres de relation client") are services dedicated to handling inbound requests from customers and/or prospects, made through one or more communication channels (phone, email, live chat, chatbot...). The term "customer relationship center" refers to a set of technical and human resources focused on customer information, after-sales service and sometimes sales.
In France, the quality of customer relationship centers is certified through the NF Service Relation Client (NF 345) certification, based on the international standards ISO 18295-1 & ISO 18295-2. This certification was created in 2004 by customer service professionals and AFNOR to give consumers a benchmark for the quality of service provided.
The FCR (first contact resolution) rate is calculated by dividing the number of calls resolved on first contact by the total number of calls handled by the contact center (or by the number of first calls if you want to exclude repeat calls from your calculation).
The FCR rate is seen by managers as one of the main success indicators for contact centers, because it reflects not only how efficiently teams handle inbound requests but also fewer low-value (repeat) calls, which leads to higher customer satisfaction and a lower cost to serve.
The churn rate (or attrition rate) is a retention indicator used to measure the scale of customer departures, although it can also apply to employees (equivalent to employee turnover). The churn rate is usually expressed as a percentage and is calculated by dividing the number of customers lost over a given period by the total number of customers over the same period (note: to calculate turnover, replace customers with employees).
CSAT (customer satisfaction) is a contact center indicator, usually expressed as a percentage and collected through dedicated surveys. CSAT is calculated as follows: the number of satisfied (4/5) and very satisfied (5/5) customers divided by the total number of customers surveyed in the CSAT survey.
Although CSAT is used in many situations, it should be analyzed alongside other performance indicators to get a representative picture of the customer experience: responses to a CSAT survey, sent by email or phone, can be influenced by the quality of the previous customer–agent interaction. For example, an unhappy customer may simply not answer the survey, whereas a satisfied customer will take the time to respond, which skews the sample.
For results that better represent your customer base, it is recommended to combine CSAT with NPS or to implement analysis of the voice of the customer.
ESAT (employee satisfaction) is an indicator used in particular by contact centers to measure employee satisfaction, expressed as a percentage and collected through dedicated surveys. Measuring ESAT is especially useful in sectors with high employee turnover/churn.
ESAT is calculated as follows: the number of satisfied (4/5) and very satisfied (5/5) employees divided by the total number of employees surveyed in the ESAT survey.
NPS, or Net Promoter Score, is a very popular indicator among customer service professionals that measures how likely a customer is to recommend a solution to people around them. A true barometer of the customer relationship and the customer experience, it is valued by professionals because it relies on a single question whose answers reveal both customer satisfaction and loyalty.
NPS is measured with a dedicated survey asking respondents to rate, on a scale from 0 to 10, how likely they are to recommend the product/service to people around them. Respondents giving 0 to 6 are considered detractors, those giving 7 or 8 are passives, and those giving 9 or 10 are promoters. NPS is calculated by subtracting the percentage of detractors from the percentage of promoters in the surveyed sample.
Like NPS, CES relies on a single question to gauge the state of the customer relationship, asking customers to rate the effort required to get their issue resolved. CES is usually measured on a 5- to 7-point agreement scale (from "strongly agree" to "strongly disagree") with a statement such as: "Resolving my issue was easy."
The CES metric is used in customer relationship management, but also to improve the customer experience (CX). CES can therefore help prevent customer churn.
CES is then calculated by subtracting the percentage of respondents who disagree (to any degree) from the percentage who agree (to any degree) with the statement.
QOS (quality of service) measures the level of service quality delivered by the contact center to its callers. QOS is often measured with a combination of individual metrics such as the FCR rate, the volume of calls handled daily or the average call handling time (ACL).
QOS therefore makes it possible to define, with objective measures, the level of service an outsourcer must deliver, and it is a key topic when drafting the SLA (service level agreement). While the outsourcer delivers the service, quality monitoring processes (which can now be automated with artificial intelligence) make it possible to verify that the contractual QOS terms are being met.
The SLA (service level agreement) is the contract describing the level of service a provider delivers to a customer. It identifies and quantifies the characteristics and metrics of the service to be delivered in order to guarantee its quality to the customer. For contact centers, for example, a very common SLA clause is a commitment to answer 100% of inbound calls within X seconds (agreed between the two parties).
The SLA is particularly important for companies that outsource their customer service, especially to call centers. Among other things, it guarantees consistent levels of responsiveness, customer care and customer satisfaction across the company's entire customer relationship, regardless of the channel used to handle complaints.
ASA (average speed of answer) is a crucial metric for contact centers' operational efficiency and their customer experience. It measures how long it takes an agent to answer an inbound call, from the moment the customer is put on hold until an agent picks up to handle the request.
If it is too high, ASA can indicate a lack of efficiency in agent teams. New approaches (digital tools, management, internal organization...) can then be put in place to reverse the trend.
ACL (average call length) is another crucial metric for contact centers and their operational efficiency. It measures how long it takes an agent to handle an inbound call, from the moment the customer is connected to the agent until they hang up after the request has been handled. A high ACL can indicate a lack of efficiency in agent teams, which hurts the customer experience. It is therefore useful to set an ACL target so teams can assess and manage the workload across the floor.
To calculate it, divide the total duration in seconds of all inbound calls by the number of inbound calls: the result is your contact center's ACL in seconds.
CPC (cost per call) is a fundamental indicator of your contact center's operational efficiency. It is calculated by dividing the total cost of all calls handled by the total number of calls, and its average value varies widely across industries.
Many contact centers set a CPC target and steer their customer service improvement plans toward lowering it (even more so for contact centers with sales targets). However, CPC should not be used as the only indicator, as a very low CPC can reflect low investment in customer relationship management and poor service quality. It is therefore essential to use it alongside other indicators, whether related to cost efficiency or agent efficiency.
The call abandonment rate, like CES or FCR, is one of the indicators of the quality of the customer experience, used to assess how painful resolution is by measuring the number of customers who were put on hold and then hung up before being served. This indicator is critical to minimize end-customer frustration, although it varies by context: a customer calling for technical support will on average tolerate waiting longer than a customer contacted with a sales offer. A call abandonment rate of around 5% is generally considered acceptable for a contact center.
The call abandonment rate is usually expressed as a percentage and calculated as follows: subtract the total number of calls answered from the total number of inbound calls, then divide the result by the total number of inbound calls.
The channel mix is an indicator measuring which contact channel customers use to reach an agent. Modern contact centers do far more than handle inbound calls and must also manage requests made via SMS, website chat and social media. The channel mix shows where inbound requests come from, so you can adapt your contact center to the main types of inbound traffic.
Tracking the channel mix over several years has revealed a general trend: voice channels such as call centers are declining in importance (still heavily used by large groups but far less present in small businesses), while "self-service" support channels such as chatbots are booming. But the decline of voice channels does not necessarily mean less use of the voice of the customer (VoC), whose semantic analysis principles can also be applied to "self-service" channels.
Speech analytics validates, at scale, the compliance of interactions against the quality scorecard defined by the contact center. It also helps identify best practices that feed into agent training.
On the customer side, speech analytics prevents churn by identifying at-risk behaviors. On the agent side, it supports upskilling and ensures impartial evaluation.
Speech analytics identifies the keys to success and the obstacles to sales, so you can then spot cross-selling opportunities or adapt existing offers.
Customer pain points are identified more easily and escalated to the relevant teams, putting user feedback right at the center of the customer relationship improvement process.
Planning a speech analytics project?
See what Batvoice reveals about your own calls, with a quality expert.
Setup, results, technology and GDPR: the essentials before you get started.
