There is much that gets said and circulated about Speech Analytics, and not all of it is accurate. Even those who are very familiar with the technology can harbor misconceptions about the logistics and requirements. Without the help of a professional service, it can often be a disorienting, or uphill, process. Though we can’t quite walk you through all of it in this post, we can give you some pointers about common mistakes and outdated ideas.
Let’s begin with the attitude that speech analytics is a set-it-up-and-walk-away kind of solution. Optimizing your speech analytics software tool so it is working for you by collecting useful data is a great start, as is a good knowledge of your keywords and phrases. However, to understand what is being said and how and why it matters is a different story. The interpretation of the data takes experience and advanced methodologies. In this process, there is no substitute for the aid of a senior speech engineer from a credible professional service.
The second misconception is that SA is a definitive answer, rather than a toolbox. Instead of expecting speech analytics to be a panacea, think of it as a handy assistant pointing you in the right direction and helping you spot patterns.
Thirdly, there is the myth than any manager with rudimentary skills can oversee the speech analytics process. While theoretically, this is true, the results it will yield will be disappointing and incomplete. To harness the potential of SA, one needs to follow through consistently in a variety of ways. There are three major areas of upkeep – administration, business analysis, and interactions monitoring, and ideally no one person will be responsible for all of them.
The “administrator” will have to implement the speech analytics tool and handle all server and database connections. The “business analyst” will be in charge of carrying out all the analytical tasks that are essential to gather, interpret, and share findings with higher-ups so that they are coherent and compelling. Finally, the “interactions monitoring analyst” will listen and go through specific calls based on recommendations by the business analyst. There will have to be a persistent and continual commute between the harvested data and the practices of the business, one feeding into the other, while maintaining communication between the different departments.
The next misconception is two sided – one is the belief that a successful project can be designed and executed without listening to recordings of real calls, and the other is that only what the agent has said is worth listening to. The raw data of the calls can feel cumbersome to navigate, but it is really helpful in establishing a real world benchmark. Then you can hear not only what is said, but how, as well as get a better grasp of the context. The best way to hear the “voice of the customer” is to actually listen. And though analyzing the agent’s side of the conversation can yield substantial insights regarding employee performance, challenges, and script utilization, it is woefully inadequate for acquiring a full picture of the interaction’s dynamics.
We will leave you with two final tips. 1) Audit your data. Commit yourself to the trial and error of auditing search results until the speech engine is fully aligned with your business goals. 2) You don’t have to do it alone. Check out KOVA Corp’s Verint Media Recorder Text Analytics for Customer Satisfaction which functions as a supplement, and at times, substitute, for SA. It will help you cast a wide net and sift through multiple channels to find the most constructive feedback from your network of customers.
