Companies face challenges in keeping up with customers’ use of more channels of communication, their increasing expectations for service, and the huge volumes and many types of customer data they are now generating. They also need to understand the behavior and sentiments of customers and of their own agents. The latest version of the Nexidia product, Interaction Analytics 11, that was announced has been designed to meet these challenges. At the heart of the product is the ability to extract information and insights through speech analytics, which Nexidia has focused on since its inception.
In general speech analytics comes in two basic forms: automatic speech recognition (ASR) and phonetic speech analysis. There are some differences in the two. ASR essentially searches through audio recordings to spot words previously specified by users. To prepare to use it, a company must create a dictionary of words its users want to find; then they run the software against the audio source (typically call recordings), and it produces an analysis showing where and how often it found the words, the contexts in which it found them and trends such as a word appearing more often than in previous analyses. To change the words being looked for, users have to change the dictionary and repeat the process. The same applies to working in another language: Users create another dictionary and run the process. In the fast-changing world of customer engagement, such a process can be too slow and cumbersome to keep up with business demands.
In contrast, phonetic analysis, which Nexidia uses, doesn’t require a dictionary. The software runs against any audio source and creates a phonetic index that is time-stamped. Users create structured queries that the software uses to find and analyze a raw audio source, and produce the required reports and analysis in much the same way as ASR does. In this case users can concentrate on building queries, and new ones can be run against the same audio source; it requires no new dictionary and works in any language by changing the language pack included with the product.
Interaction Analytics 11 takes this process further, into what Nexidia calls neural phonetic speech analysis. In addition to audio, the software can process text data from sources such as email, surveys, chat scripts and text messages. The system uncovers information from these combined sources and applies the analysis, thus giving users a fuller picture of customer interactions. The system can also use predefined rules to uncover agent and customer sentiment, adding another dimension to the analysis. The outputs from the initial discovery phase and sentiment analysis can be run against prebuilt models concerning business issues such as customer churn and retention or sales effectiveness. All of these outputs can be included in enhanced reports and analysis, which can be extended to include key metrics.
As well as adding these functional capabilities, Nexidia has rearchitected the product to be more scalable and able to process data
Ventana Research believes that to derive full benefit from any application, especially analytics, it should be used within an overall performance management framework. This should include three steps: Understand, Optimize and Align. In this context, Understand uses batch and real-time analytics to show what has happened and what is happening; Optimize uses these outputs to decide which changes are required; Align creates an action plan to ensure the changes are brought to bear. The analytic process uses metrics and is cyclic so that repeating the cycle shows the impacts of changes and other changes needed going forward. Nexidia supports a similar approach, which uses its discovery tools to reveal what has happened and what is happening, root-cause analysis to understand why it happened and a metrics-driven approach to improvement. For example, companies can enhance quality management by analyzing more sources of data (such as surveys, customer feedback and call recordings), understanding drivers of agent and customer satisfaction, and fine-tuning coaching and training to improve the customer experience. My research into the agent desktop and customer service shows that very satisfied agents twice as often as those less satisfied deliver on key customer-related metrics such as customer satisfaction, net promoter scores and first-time interaction resolution.
Nexidia’s products give users insights into the processes connected with improving the customer experience. I recommend that organizations examine how Nexidia can help improve the outcomes of customer interactions using its next generation customer analytics called Interaction Analytics.
Regards,
Richard J. Snow
VP & Research Director