You call a company and hear the automated message: "This call may be monitored for quality assurance." You're soon in a fit of rage because you've been on hold for too long, your package was shipped to the wrong address, and now some kid in Nova Scotia is playing your Nintendo Switch. You don't care whether or not the call will be monitored; you just care that you receive the customer service necessary to rectify your situation. You probably realize your call will be recorded, analyzed, and used to teach service agents how to better manage customer relationships (not that you really care). But what you probably don't realize is, there is software running alongside this call, measuring the timbre of your voice, tracking keywords such as "angry" or "manager" in order to guide the representative to deliver a better resolution to you.
Continue Reading Below
What Is Speech Analytics?
The speech analytics market is expected to reach $1.6 billion by 2020, according to a MarketsandMarkets report. The industry is composed of companies that offer basic services such as recording, transcribing, and providing businesses with analysis of historical calls. You'll also find real-time engines that can send alerts to supervisors during calls, alerting them that their new service rep has angered a long-time customer. This kind of real-time speech analytics is designed to automate the call-monitoring process in order to improve customer service, as well as to provide marketing and sales insight.
"The historical piece has primarily been used to improve performance in the contact center," said Ian Jacobs, Principal Analyst at Forrester Research. "You can find out how agents are doing. Companies are also using this to train salesforces. Listening to those calls and using that to help train the salespeople to be better at closing sales or getting leads."
Today, more than 35 percent of companies with at least a 50-person contact center have implemented a speech analytics tool that can use a real-time decision-making engine to push offers to contact center agents, according to Forrester. So, if a customer is calling to say that he is unhappy with his tractor, then the system will pick up on his dissatisfaction. The agent can then push an offer to upgrade a warranty or buy another tractor, entirely depending on what the specific situation merits.
Adoption of real-time speech analytics tools is expected to increase but there are great barriers to entry, especially for small-market companies that don't have the vast call center infrastructure required to make the best use of these tools. For example, the most cutting-edge tools are still too advanced for the traditional call center supervisor and his or her employees. Additionally, because of a business' ever-evolving product and service line, speech analytics tools will have to be adjusted and optimized on a regular basis to provide the most value to the company. And let's not forget the most important factor: cost. Speech analytics tools, like most other forms of enterprise software, can be quite expensive. Because companies have traditionally used these tools for training purposes, it has historically been difficult to prove good return-on-investment (ROI).
Speech Analytics for Marketing and Sales
This brings us to the next generation of speech analytics tools: those that can be used to monitor sales professionals and marketers in order to push offers in real time. Although the telephone may no longer be the go-to device for brand interactions, driving, tracking, and optimizing phone calls influences more than $1 trillion in US consumer spending each year, according to a BIA/Kelsey report. This is a golden opportunity for the speech analytics industry to prove its worth; the opportunity is reinforced by the fact that every dollar earned during a monitored call can be traced back to the initial software investment.
"Because these [real-time speech analytics tools] are expensive, most early use cases were around revenue generation," said Jacobs. "It's a lot easier to justify the purchase if it's generating revenue. That's where real-time has been starting."
One of the companies leading the speech analytics ROI revolution is Marchex, a speech analytics tool that is specifically designed to help businesses determine how digital advertising and marketing affects purchases and satisfaction. Essentially, Marchex's software monitors calls in real time and for posterity. The software listens for keywords that pertain to website layout, video ads, banners, email marketing efforts, and anything else that relates to marketing. If customers are calling to complain about your website's design, then you can immediately adjust your website to better facilitate navigation and purchases. If people love your new YouTube video and they're referencing it during service and purchase calls, then you can use that data to determine how much to spend to amplify video traffic.
"We can ask the same questions a call center may ask but it turns into an interesting tool for marketers," said Guy Weismantel, Executive VP and CMO at Marchex. "More and more, with the advent of the CMO having digital information about prospective customers, you're dropping cookies on them, you're re-targeting them. Marketing has gone from the brand steward to one that's focused on driving revenue."
The Major Players
In addition to Marchex, which admittedly focuses on the revenue-generation slice of the speech analytics market, there are companies such as Nice and Verint—the Coke and Pepsi of speech analytics, respectively. These Israeli-based companies are "bitter rivals" according to Jacobs, but he said they're both excellent options for anyone considering a move into call analysis.
Although the majority of Nice customers are primarily focused on post-call analytics, the company is emphasizing its real-time functionality to improve customer analytics, engagement, and operations workflow.
Conversely, Verint wants to be the speech analytics tool that you plug into your overall workforce optimization tool. In this scenario, Verint's software would catch that an agent made an error on three successive calls. Verint knows that agents who make three errors are automatically required to repeat their training, so the system sends an automated message to a supervisor, scheduling training for the agent. The rep will, from within Verint, be able to schedule a training session that will take place on an integrated online learning platform.