7 Speech Analytics Mistakes That Can Cost You Big


Finding the right Speech Analytics Solution starts with knowing the right questions to ask. It’s also important to know that avoiding Speech Analytics mistakes requires you to know the questions not worth asking.

Thanks to Roger Lee of Gridspace for sharing his insight at Customer Contact Week. Roger is a long-time contact center industry veteran and a respected thought-leader. At the event, he conducted a workshop about Speech Analytics, and he graciously shared with me his observations about what occurred during that workshop.

Here is an edited transcript of my interview with Roger:

Jim Rembach: This is Jim with the Fast Leader Show and Call Center Coach, and I’m here at CCW (Customer Contact Week)  with my good friend Roger Lee of Gridspace. Now, Roger, you had the opportunity to sit on some interactive educational sessions, and you were facilitating actionable analytics sessions. However, a lot of the questions that you were fielding and we’re helping people with were in regards to speech analytics, so what were some of the things that people are asking about.

Speech Analytics Mistake #1: Being driven by misconceptions

Roger Lee: Yeah, thanks, Jim, great question. So, about ninety percent (90%) of the audience that attended the sessions were new on looking for a speech analytics platform. And so, some of those questions they were asking were interesting because I think there is a large amount of maybe misconception around what speech analytics is exes and really what it can do for the organization.

Mistake #2: Not knowing what you’re solving for

One of the questions I had asked was, so what is the business problem you’re trying to solve a speech analytics? About fifty percent (50%) of the attendees really did not know how to answer that question.

Mistake #3: You should know WHY people are calling

They would say I want to use speech analytics to help me understand why people are calling. Well, you’re call center or contact center you should know why people are calling right. So, it was more of validating why people are calling, then that’s a good use of speech analytics.

Mistake #4: Improperly selling it to the C-suite

Another question was, how do I sell it to the C-suite. Awe, that’s a really long discussion, but I will tell you there are definitely elements of a business case that we have shown can help as you are designing and developing why speech analytics is so important to your organization. And it can be very simple, but I look at two particular areas of a business case, improving agent performance and also how to improve the enterprise as a whole. So, there are two pieces. Some of the initial conversations in the workshop was about agent performance, how to streamline the QA (Quality Assurance) process. All very important, but that necessarily won’t sell it to the C-suite.

Mistake #5: Not testing transcription accuracy

So, if you’re looking for speech analytics platform to help you address your business problems within your contact center, here are some things that you should be aware of.

One, take a look at transcription accuracy. How accurate is the information that you’re gleaning from those call recordings or live conversations?

Mistake #6: Not testing for user-friendliness

Two, how easy is it to use the application from an end-user perspective? Do you need a number of resources in addition to your already tasked team of QAs or analysts to review calls and to use the application?

Mistake #7: Not knowing the back of the product

Third, when you’re looking at the technology, take a look at how is the back of the solution being used. In other words, is the speech analytics solution phonetic-based or semantic-based? There are definitely clear differences on how the tool recognizes and transcribes and uses the information from the call recordings.

Jim Rembach: I think one of the things that’s also important, you and I had talked about before, is really understanding the difference between intent and emotion. Tell us a little bit about that.

Roger Lee: Yeah, that’s a great call out. There’s a number of conversations or questions regarding sentiment right. So, they want to understand the caller’s intent or sentiment of the call. Or even how the agent responds.

I use the term more of what I call emotion modeling. That is a big difference. Sentiment is really looking at the words; emotion modeling is looking at the tonality of the conversation in addition to the context of the conversation. Those two attributes will make up the emotion modeling will actually give you a fairer or clear representation of what is happening.

Jim Rembach: Well, I think the biggest difference is not what the intention is, or desire is, heck even as far as the resolution, but it’s what happens from an emotional perspective. Because that’s going to affect how they feel about the organization when the call is concluded.

Roger Lee: Exactly. Even during the call, there’s an opportunity. If the emotion is not where it needs to be, there’s an opportunity for the agent along with a supervisor to try to save or change that emotion. So, at the beginning of the call, it could be negative, but by the middle of the call, you could be neutral, and hopefully, by the end, it could be very positive. There’s a great opportunity there to leverage the emotion piece accordingly.

Jim Rembach: Roger, thanks for sharing your knowledge and wisdom and helping us all make better decisions when it comes to avoiding Speech Analytics mistakes.

Roger Lee: Thank you very much for your time.

Please Share

  • [VIDEO] Seven Speech Analytics Mistakes That Can Cost You Big – Click to Tweet
  • What are the Speech Analytics questions NOT worth asking? – Click to Tweet
  • Going Beyond Speech Analytics 101: These are the Questions to Ask – Click to Tweet

Additional Resources

Published with permission from the original at Call Center Coach.

Leave a Comment