Speech technologies have tremendous potential in the contact centre, given that 92% of all business and customer interactions still rely on voice. As we noted in previous articles, extracting CX insights using analytics is a top priority for 75% of companies, and a big part of this endeavour is speech analytics implementation. There are X best practices that can help you make the most out of speech analytics at your organisation – but before we delve deeper, let’s quickly explore how the technology works.
What is Speech Analytics and How Does it Work?
Speech analytics is a set of software algorithms that can extract the unstructured data trapped in calls or any other voice communication and convert it into meaningful insights. It captures all your contact centre interactions and their metadata to craft transcriptions as well as arrive at key acoustic measurements. This gives you valuable insights such as:
- Adherence to scripts and specific keywords
- Agent tone of voice and attitude
- Call clarity is reflected in word tempo, agitation, and silence
- Real-time dispute detection
According to 2020 research, implementing speech analytics to achieve these (and other) use cases is part of the immediate agenda for 15% of companies. Here’s how you can make the most of this investment.
5 Best Practices for Implementing Speech Analytics
1. Identify use cases before you launch – Speech analytics performs the essential but cumbersome task of converting uttered, unstructured speech into legible transcriptions and actionable insights. This has a variety of applications, from training your workforce to understanding customer sentiment, from identifying product trends to real-time quality assurance
2. Combine with text analytics – In addition to pure-play speech analysis, it can be helpful to pass other interactions (email, chats, surveys, social media, text) through similar algorithms. This will give you a holistic picture of the customer and provide a context to every interaction, thereby making the speech analysis more accurate
3. Start with a limited pilot with a clear KPI – It can be tempting to initiate a large-scale speech analytics implementation at a greenfield state and try to unlock its many broad benefits across the organisation. But this makes it difficult to get leadership buy-in, not to mention challenges in tracking progress and ironing out software issues. Instead, start with a targeted pilot (e.g., detecting long periods of silence that’s bringing down FTR), and measure KPI-based improvements (i.e., FTR boost after training based on speech analytics findings)
4. Consider partnering with a best in category provider – Several 360-degree contact centre solutions ship with speech analytics capabilities, but these may not always be best-in-class. It might be a good idea to explore specialised speech analytics companies such as CallMiner or VoiceBase and integrate with your existing systems
5. Focus on training and change management – speech analytics entails a fundamental change in how agents work. In addition to manually keeping a check on call quality, they must now refer to real-time speech analytics and look for keyword suggestions/misses, prolonged silence, etc. make sure to train the workforce before implementing and weave the technology into your process flow
