Speech Analytics: Three Things to Know Before You Deploy
The industry is abuzz with the opportunities presented through speech analytics, but most call centers are still unsure of how to utilize the provided data and convert it to an action plan. For more understanding of the history and opportunity presented with deploying a speech analytics system in a contact center, we called on Craig Reines, Vice President of Corporate Development and Strategy at CallMiner, a new partner of Spoken Communications.
Reines stresses that a properly deployed speech analytics system should be used for analyzing interactions across all customer contact channels and producing actionable insight.
The biggest miss is thinking that these tools are only used for audio word spotting. We should really refer to the system with the broader term of “contact analytics” to acknowledge the multichannel approach to analysis.
A history of analytics
In 2001, Reines was part of a team testing pilots of speech analytics, and at the time, he says, it didn’t really work. Not because it wasn’t a great concept. Back then, there were computational power and bandwidth limitations that made processing large quantities of audio extremely expensive. As a result, in 2004 there were only 25 known implementations of the technology in place.
However, as storage and bandwidth costs decreased, speech analytics advanced. In the last five years, contact analytics has evolved even further from those 25 backroom projects to becoming critical to supporting operations. By 2012, the number of implementations jumped to 3,170 and now, research suggests the speech analytics market, currently valued at $214 million, is projected to grow by 20 percent this year.
Why is it growing so much? Because once lowered technology and processing costs made speech analytics more accessible, businesses could easily see the power of the ability to marry data between call interaction, customer surveys and website visits. When the deployed system goes through and analyzes each point of contact from customers with the business, businesses are able to get a more accurate read of needs, concerns and opportunities.
Manual vs automated analysis
In a post on the CallMiner blog, Reines shows a great example of a miss in customer experience due to traditional quality analytics (QA) system. The customer went around in circles with multiple agents on multiple calls, but chose not to get angry or use words that would trigger the call to be put on alert. As a result, the customer left dissatisfied with his experience, and the company had no idea.
However, a challenge with the QA approach to speech analytics is that there are many misses in QA systems where supervisors monitor a random selection of five calls per agent.
It is a rather manual process that is not terribly accurate and is pretty subjective.There is a very small chance that the above calls would have been selected for one of the agent’s three monitored calls.
A properly deployed contact analytics system would have been able to see that the customer had called multiple times and found the trend in the calls, which would have empowered the agents to change this customer experience. Recording and analyzing objectively with an automated system takes subjectivity and filters taken out of the process. As a result, businesses are able to truly see that customers are leaving satisfied with their experience.
Bonus: actionable data for agent improvement
The system also creates an opportunity for agent growth and improvement by giving them access to all of the data available on their performances. Strict scorecards that do not provide people with the opportunity to contextualize conversations can leave agents feeling frustrated. While it is up to each business to decide who has access to available data, bringing timely and actionable feedback to your front line customer service representatives will generate marked improvement.
Know before you deploy
So what does your business need in order to deploy a successful contact analytics system?
- No silver bullets Understand that the system is not a magic wand, and deploying the system will not provide results with the push of a button. The system will automate the data collection process then identify and contextualize data points for QA managers and supervisors to look at and take action on.
- A holistic approach The old scorecards are gone. The new, more comprehensive, data available challenges call center managers and agents to take the data and work with agents holistically to see how they can improve troublesome areas.
- Make a plan A strategized plan is essential to deployment. Determine in advance what metrics you’re looking for and what your end goal will be. Better customer experience? Increased sales? How does this tie into the greater company goals?
As the use of these systems grows, contact centers are in store for substantially improved feedback within all levels of the organization, new use cases for real time contact analytics and enhanced marketing campaign management.