Financial services firms record customer calls to meet their regulatory obligations: know your customer (KYC), anti-money laundering (AML) processes and Consumer Duty, to name a few. These call records can give valuable insight into how customer support operatives are identifying potentially vulnerable customers and any key problem areas for consumers. Your third line of defence must review these calls to uphold quality, maintain regulatory compliance and demonstrate good customer outcomes.
But this is a monumental feat, with larger companies handling thousands of calls a day, many lasting for more than half an hour. This has a significant impact on the proportion of calls your review teams can assess, with a knock-on impact on the degree of assurance you can offer to senior management. Voice analytics can help simplify that process.
How does voice analytics work?
Voice analytics uses voice-to-text transcription, which automatically converts spoken language into written text. The process involves several stages, starting with recording the audio and ending with data analytics to gain insights from the resulting text.
First off, you can apply automatic speech recognition (ASR) software to process your call-centre recordings and convert them from spoken audio into text. Machine learning is used to train the ASR software to recognise and interpret different accents, dialects, and speech patterns. This allows the software to adapt to the speech of individual speakers and improve the accuracy of the transcription.
Once you’ve transcribed the speech, you can use data analytics techniques to see what your customers are saying and the general sentiment they’re expressing. Sentiment analysis identifies the emotions or opinions your customers express within the call transcript, and can tell you if they're happy, angry, frustrated, or satisfied. You can also assess whether the call follows the right KYC processes for customer identification, and recognise potentially vulnerable customers who may require additional support.
Your first-line, second-line, and third-line teams can use these insights to improve service delivery and call centre performance.
Embracing continuous monitoring
Voice analytics can feed into a near-real-time dashboard, which you can share with your first-line, second-line, and third-line teams to address issues sooner. Your risk management team can set trigger conditions for automated alerts of any outliers, and red flags for immediate attention. This will ultimately help reduce issues due to regulatory breaches, improve customer satisfaction, and reduce the potential for reputational damage.
You get a clearer picture of the whole customer journey under Consumer Duty. Sentiment analysis will help identify parts of the journey where customers are feeling frustrated or angry, giving you an opportunity to intervene and act towards better customer outcomes.
Reducing pressure on resources
Voice analytics can reduce pressure on your resources and allow your team to review more calls. With a basic assessment of 100% of your calls, rather than a sample, your team can focus their efforts where it really counts. This gives stakeholders greater assurance over your regulatory compliance and conduct, with a stronger audit trail to evidence it. It’s good news from a people perspective too. Less time on repetitive, manual tasks means they can focus on more challenging work that’s more fulfilling, making them more likely to grow in their current role.
Of course, voice analytics isn’t perfect. One of the biggest challenges is the strength of the transcription technology, which can struggle with accents, speech difficulties, or background noise. But the technology is continuously improving and the current benefits outweigh the downsides of false positives.
A four-step process for implementing voice analytics
To get started with voice analytics, you can think about four key areas.
Step one: discovery
What do you want to use voice analytics for? What do you want to measure? For example, you might want to look at call sentiment or themes in complaints across different customer age ranges, over a given period.
Step two: qualification
Are you currently collecting the right data to support your discovery goals? For example, are you collecting data on customer age and how long do you keep call centre recordings? You may need to retrospectively collect dates of birth (and factor that into data collection processes moving forward) and focus your initial time frame to match your longest-held recordings.
Step three: exploration
Run a series of data analytics steps to identify key trends and test any working hypotheses as to how particular customer age groups may influence the complaints or themes within their calls.
Step four: realisation
How can you continue to capture that information moving forward? What additional information could you capture that would add value and create further insight? Putting the right operating model in place to capture, analyse and (most importantly) react to insights from the data is crucial.
As you get a better understanding of your customers’ needs over time, you may run through these steps countless times with varying search criteria.
Getting started with voice analytics
Firms across the financial services sector have a wealth of data to draw from, and it’s essential to use that information wisely to maximise the business insights available. Voice analytics can be transformative and add significant value to your organisation, but it can be difficult to know where to start and how to develop use cases for it. Building the necessary skill sets, infrastructure and processes will take time, so it’s important to get a head start as soon as possible to stay on top of regulatory compliance and meet customer expectations.
