One would think that with all the conversations about chatbots, social, and increasingly video channels, on text and visual communications, that speech, yes voice, is becoming passe for contact centers.
But that appears not to be the case, based on recent industry research. In fact, the role of speech, in particular speech technology, continues to grow as organizations seek deeper insights from customer engagements and experiences.
We had the opportunity to discuss these trends and findings with Shadi Baqleh, COO at Deepgram. Here is our interview.
Q. What are the top three opportunities for speech technology in customer service, support, billing, and sales? And what are their drivers?
This year’s State of Voice Technology report found that the top three opportunities for implementing speech technology were: improving employee/agent productivity (87%), identifying new business opportunities (77%), and increasing revenues (62%).
Customer loyalty is harder to secure than ever, leading more and more companies to differentiate on customer experience (CX).
Speech technology is seen, then, as a key piece of CX with 73% of respondents noting CX analysis as the most impactful use for speech technology, followed by conversational AI (artificial intelligence) at 54%. Meanwhile, 75% report they plan to increase their speech technology budget in 2022.
Getting into more detail, increasing employee/agent productivity can be done with voicebots and automated support enablement to identify solutions to callers’ issues.
If you can transcribe all your audio and video customer interactions, you may find insights on new products or services with AI searches for pain points.
One last use case we have seen a lot is in sales enablement, recording all sales calls to coach, recommending next steps, and recommending add-on sales or products.
Q. What functions are contact centers using speech for, and why? And by the same token what functions can they use it for but are not, or in lesser numbers, and why?
Call center customer interactions contain a goldmine of valuable information, and there are several technology solutions that allow call centers to make proper, actionable use of their data. Speech recognition technology is one of them.
Speech technologies are invaluable when it comes to recognizing a huge amount of vocabulary across many languages, identifying relevant words and phrases, analyzing the tone of voice, and even estimating the age of callers all from interactions with call centers.
As sentiment analysis begins to take a stronger hold, this may be integrated into call center interactions as well, elevating the level of customer service provided based on mood detection.
Call centers can use insights driven by speech recognition technology to improve various aspects of their business operations as well.
The insights can help identify customers who need special attention, pinpoint industry trends, and analyze behavior and purchasing patterns. Overall, call center speech technology can function to improve personalization for the CX.
Training Agents With Speech?
Speech technology is an excellent tool, when implemented right, to enable an excellent CX while reducing costs. But can it help contact centers train their agents?
Here is what Shadi Baqleh, COO, Deepgram, had to say in response:
“Speech technology is being used sparingly in training contact center agents currently: mostly to listen for specific compliance issues, for quality assurance, or confirmation of an agreement. What audio to review may come from ad-hoc customer or manager complaints.
“But this method misses larger market patterns, issues, and needs of their customer base that can add value to all contact center agents.
“What can we do now with 90%-plus accuracy, real-time streaming with millisecond lags, 300X speed on transcriptions, and lower costs that are now possible on the speech to text? These new abilities allow contact centers to do so much more.
- For onboarding, new agents can practice their responses or sales pitches and using AI speech technology get feedback on their customer interaction performance before they get on an actual call.
- Shorten onboarding time and increase speed to productivity.
- Transcribe and analyze all customer calls to find larger product or service issues that can be solved globally and added to the agent’s training, thereby reducing time to resolution.
- Provide real-time enablement to agents with on-screen prompts. If the caller sounds like they are about to churn, talks about an issue that is in the knowledge base, or sounds like they might be open for an upsell, speech technology with an AI knowledge base provides on-screen assistance to the agent, thereby reducing churn and increasing add on revenues.
“The challenges are that contact centers are still seen mostly as a cost center. Growth trumps cost savings, so adding new tools to help with agent training is hard to justify.
“However, new abilities can truly make a contact center a revenue growth center. If you can reduce churn and increase upsell or add-ons with better tools for your agents, you are now a revenue driver.”
Q. Conversely, what are the top three challenges to applying speech technology and their causes?
The top three challenges in using speech technology are accuracy, costs, and speed of transcriptions.
1. Accuracy. With speech technology, poor accuracy leads to poor understanding. Poor accuracy can be driven by poor audio quality, background noise, accents, dialects, and work-related jargon. Speech models that aren’t trained to handle those scenarios.
Higher rates of accuracy are bolstered by the ability to train custom models which are tailor-fit for specific audio types, languages, accents, and environments.
This is why deep learning is more important than ever in powering meaningful voice experiences. Deep learning enables companies to scale massively and glean advanced analytics at costs, speeds, and accuracies needed for the modern contact center. We do know that with it accuracy rates of 90%-plus and more can be achieved and improved with additional training.
2. Costs. In the past, when companies began implementing voice technology as part of their customer service solutions, transcription was very expensive and slow, so contact centers would only transcribe a sample of their audio data due to cost.
3. Speed of transcriptions. Because the vast amount of audio many companies need to have transcribed, they often cannot have one day’s audio transcribed within that day.
Q. Isn’t the use of voice for customer service stagnating or declining in the face of a plethora of expanding and evolving text-based and video interaction channels?
Voice is the most intuitive form of communication, and while we are seeing advancements in text-based and video technologies, they will not diminish the important role voice plays. From our survey efforts, an overwhelming 99% of respondents said they view voice-enabled experiences as a critical part of their company’s future enterprise strategy.
Customers call when they are faced with the toughest questions or issues because it is still a faster form of communication than typing. Most critically, calling conveys a customer’s frustrations, which is so important to gauge overall customer satisfaction.
As aforementioned, sentiment analysis from conversational AI will be able to personalize the CX for call center users like never before.
…while we are seeing advancements in text-based and video technologies, they will not diminish the important role voice plays.
Going back to the frustrated customer, they may need immediate attention or feedback. Sentiment analysis technology will be able to determine next steps much better than a text-only operator could.
Additionally, there are a number of verticals, such as the healthcare and financial services industries, that are increasingly integrating speech technology into their daily operations as it works best for their work needs.
Employees in these industries have busy schedules and using voice technology to lessen the time spent on administrative tasks and general questioning will add valuable time back into their days.
Q. Are customers becoming or less accepting of speech technology?
As a society we are becoming more accepting of speech technology or “this call may be recorded for quality purposes” announcements.
…77% of respondents think that mass implementation of voice technology will occur within the next five years, clearly representing that there is an appetite for growth of adoption of [that] technology.
The report revealed that more than half (64%) of respondents expect speech tech to be one of the most important aspects of their future enterprise strategies. While this number is lower than 2021, which was 85%, we believe that difference is stemming from consumers and businesses not feeling as pressured by the COVID-19 pandemic, and in-person customer service interactions becoming more comfortable.
At the same time, the report found that 77% of respondents think that mass implementation of voice technology will occur within the next five years, clearly representing that there is an appetite for growth of adoption of [that] technology.
Q. What are your recommendations for contact centers to maximize their use of and the benefits from speech technology?
One recommendation for call centers is not to ignore the plethora of valuable unstructured data that is available to them through voice data.
Historically, call centers have taken a “store it and forget it” approach to their customer call data, leaving valuable customer insights forgotten in a database.
Instead, call centers can tap into valuable unstructured data to pull out key terms and unearth valuable customer insights to better serve the customer and improve CX: versus leaving the data to sit without any kind of analysis.
Another recommendation is to seek out newer providers of contact center technologies, SaaS-based CCaaS (contact center as a service), call analytics, sales enablement, or support enablement companies. You may find that they are cheaper, faster, and better than your current system and easier to implement than you think.
These newer technologies are also focused on easy-to-connect APIs and great user interfaces, which eases implementation and increases end-user acceptance.
