12
Voice analytics
What is it?
Voice analytics, also known as speech analytics, is the process of extracting information, meaning and insights from audio recordings of conversations.
This form of analytics can analyse the topics or actual words and phrases being discussed in a conversation. This can be extremely useful for security purposes and certainly counter terrorist units inside most governments monitor a lot more of our conversation than we might imagine in order to identify people talking about things that they either shouldn’t be talking about or to help identify potential treats.
Voice analytics can also be used to analyse the emotional content of the conversation above and beyond the actual words and phrases being discussed. When we get angry, for example, the pitch and tone of our voice changes, this is also true for most of us when we lie and so voice analytics can identify these changes in emotional state.
When do I use it?
All businesses need to keep their customers happy if they want to stay in business and stay ahead of the competition. If you have a product or service that requires technical assistance or you have large customer service call centres then this type of analytics can be really useful in maintaining and building ongoing customer relationships as well as highlight issues that need to be addressed.
You could, for example, use voice analytics to help identify recurring themes around customer complaints or recurring technical issues. These insights could help you to spot these potential pitfalls quicker and solve them before your customers take to social media to complain.
Voice analytics can also be used to help you identify when your customers are getting upset. By analysing the pitch and intonations of conversations taking place in your call centre you can gauge the emotional state of your customers and intervene earlier when they are getting angry or frustrated. The amount of speech and location of speech versus silence, i.e. call hold times or periods of silence, can also help customer-facing businesses provide better service and keep their customers happier. As a result, the conversations we’ve been told ‘may be recorded for training purposes’ can actually be used for training and provide useful insights instead of being lost or recorded over.
This type of analytics is also very useful in helping to identify underperforming customer service representatives so they can receive additional training or coaching, and can automatically monitor the level of customer service provided on calls, i.e. does the emotional customer end the call still upset or are they calmer and much happier? This is important information for customer retention and loyalty building.
What business questions is it helping me to answer?
Voice analytics can help you to answer:
- What do customers really think about our brand/product?
- How can we identify the customers that are upset and likely to leave?
- How can we identify lies (used in police forces and border agencies)?
- How can we make our operations more efficient (and cut down on, for example, form filling)?
How do I use it?
Voice analytic tools can spot spoken keywords, phrases or emotional tonality either as real-time alerts on live audio or as a post-processing step on recorded speech. It is actually this type of analytic ability that helps live TV and radio shows manage the unpredictability of guests. Obviously it’s important that people don’t swear on live shows, especially before the watershed, so voice analytics can help to recognise speech patterns that may be leading up to swearing and cut that person off before any damage is done.
Clearly if you want to use voice analytics you need to have voice data to analyse either as live audio or recorded conversations. And the more recent that data is the better, certainly for ongoing extraction of commercially relevant insights. The most obvious source of voice data is from call centres or customer services departments which are interacting with customers all the time.
In more cases, certainly within large businesses, this data will already exist and customers in call waiting queues are usually told that the conversation may be recorded. If this data already exists then it makes sense to use it more constructively in order to discover strategically significant information about products, processes, operational issues, areas for improvement and customer service performance. Voice analytics can provide you with information about what your customers really think about your company, products and services without much additional investment in market research. It is essentially about leveraging data you already have.
If you don’t currently record your customer conversations then you may want to start. Technology for recording conversations is very common and inexpensive and so long as you tell your customers that the calls are or may be recorded then you can use that data to extract additional value from a service that you are already providing.
If you want to know more about voice analytics and how they can be used you can explore the links at the end of this chapter. Alternatively there are commercially available voice analytics tools and providers that can help you.
Practical example
Voice analytics can be used to extract value from what’s being said and how it’s being said in a way that simply wasn’t possible a decade ago.
There are many social and commercial applications for voice analytics because it can help us to identify when someone is stressed, scared, happy, sad, or even when they are lying.
For example, voice analytics is already being used by insurance companies to help detect insurance fraud. If someone calls up a claims line and tells a representative about a new claim for a car accident or damage claim on their home then this type of analytics will help to detect who is lying and trying to commit insurance fraud. Insurance fraud is a huge ongoing problem for the industry that customers often end up paying for through increased premiums. Voice analytics is helping to identify the cheats, which will hopefully help everyone else.
As well as providing useful business insights voice analytics can be seen commercially in voice recognition software for dictaphones and smartphone dictation apps.
Plus it is this analytic capability that you are using when you talk to Siri on Apple’s iPhone or Microsoft’s Cortana which is available on new Nokia phones. By talking into your smart device the technology will decipher what you said and either take you to a specific website or remind you to send a birthday card to your friend. Also many modern cars offer a text-to-voice feature so that if you get a text message to your phone the car will convert the text to speech so you can hear your message without disrupting your driving.
Tips and traps
For voice analytics to be truly effective and useful the voice data needs to be clear and crisp so make sure you invest in quality equipment. And use the data to answer specific strategic questions and seek constant improvement rather than just seeing what the data tells you.
Recording conversations can make people nervous. For employees it can feel like a ‘big brother’ intervention that is designed to monitor what they are doing and saying. It is important therefore that any decision to record calls is positioned correctly for both the customers and the employees. Be sure to share the insights you learn with your people so they can appreciate that you are not listening to their every call but listening to what’s going on behind the call.
Plus you can’t record people without their permission so you need to tell your customers and employees that they are being recorded and allow them to opt out if they want to. Make sure you stay apprised of any changes to the law that could affect voice data and always err on the side of ethical caution.
Further reading and references
To find out more about voice analytics and how they can be used see for example: