Phone calls may seem old-fashioned compared to the newer, what we might consider “more connected” business communications channels and tools out there. But this medium is as vital to customer interactions now than it’s ever been. So, listen up. Recent advances and cost reductions in speech-to-text technology are changing the game for voice-based connections.

From sales, to customer support and internal communications, voice is still often the preferred method for the most critical aspects of remote interaction between co-workers, clients–business interactions of all kinds, really. Yet as a channel, voice is often overlooked when it comes to performing regular analytics for improvement and optimization. It’s time to look more closely at how we value our daily conversations.

Modern machines are honing human interactions

Machine learning has transformed the effectiveness of speech recognitionThe phone keeps evolving, retrofitted for a modern business era where it needs to do things like be compatible with CRMs, serve as an authentication method or spoken password, and generally to get things done that a chatbot might fry its circuits over. With the help of machines, however, voice interactions open possibilities that our rotary-phone-dialing forefathers would have marveled at.

Recording equipment and speech analytics have gotten a 21st-century makeover. Cloud-based software is replacing hardware and AI dutifully stands in for humans to extract valuable data from conversations that might have otherwise gone undocumented. Speech transcription is becoming more reliable and accurate every day, with companies like Microsoft dropping transcription error rates down to a record-setting 5.1%.

Take the Association of Public-Safety Communications Officials (APCO) International as an example of the confidence that speech analysis can instill. These folks are behind emergency calls, where the interactions between a caller and an agent can be a matter of life and death. APCO recently announced a collaboration with IBM Watson’s transcription and analytics services to improve scripted responses to unpredictable caller scenarios.

Why should we care? For a company looking to run lean and retain customers, implementing effective customer service and sales processes is essential. Yet the stats and data needed to make informed changes aren’t easy to get ahold of. You could hire staff to review and transcribe calls, or send out customer surveys to get usable numbers and/or feedback. These are, of course, expensive and time-consuming options.

Speech analytics can be used to optimize calls and processesWith speech analytics, on the other hand, you can cover all these bases in real time, without needing to invest in equipment or additional staff. AI might not be at the level where it can replace human agents in a contact center (yet), but it’s an incredible tool that can revolutionize what your employees are capable of and how your customers see you as a company.

It doesn’t look like chatbots or real-time speech recognition will be taking over our sales and customer support jobs anytime soon. But in terms of transcribing and analyzing voice interactions at a large scale, our robot friends have the edge. A task that would take a team of specialists hours to complete can be done instantly using speech analytics tools. This gives experts the ability to better focus their efforts on a pre-filtered list of calls where they can have the most impact rather than choosing a small percentage of calls to look at random.

Speech analytics solutions can even help improve sales and agent-customer interactions without invoking a supervisor’s reprimands. Post-call analysis solutions can show agents where they deviated from proven best practice scripts so they can improve next time. Some advanced systems can even do this in real time, letting the agent get back on track before it is too late.

Why investing in speech analytics is a smart move

With hardware being replaced by software-based systems, call recording is not only less expensive but also much easier. Recording capabilities can be built into your communications platform, with high-quality audio files save in and sent from the cloud. Once the investment has been made into a recording and analytics provider, the opportunities for making cost-saving adjustments to your sales and customer service processes are well worth the spend. Here are a few ways analytics can be transformed into legit savings for your business:

  • Identifying sales opportunities such as cross-selling and upselling to assess and improve agent performance
  • Flagging keywords and/or compliance issues in order to minimize the risk of fines or setbacks due to violations
  • Determining whether there is too much silence or multiple touches needed for a resolution in order to streamline support and satisfy more customers with fewer resources
  • Using insights to improve the customer service experience, which is a growing factor in customer retention, can prevent against churn
  • AI is getting better at recognizing a caller’s emotions–this can help determine customer satisfaction and is a more effective alternative to relying on customer surveys, which typically have low response rates

More effective sales agents means higher revenues and lower costsWithout speech analytics, data that could translate into major cost savings and efficiency improvements would go down the drain. According to voice analytics provider VoiceBase, companies can extract 3,700 times more information from spoken interactions between customers and agents compared to what’s written online. While an omni-channel communications strategy is becoming the standard among modern businesses, we shouldn’t leave voice communications behind when mining customer data.

Mind the gap for potential drawbacks

Voice recognition and AI-driven analytics are new and exciting technologies, but they aren’t without limitations. As I mentioned before, speech analytics are best used to enhance customer-agent interactions rather than outright replacing them. If a caller has a complex issue that needs solving or isn’t matching keywords to the letter, for example, the human brain is still way ahead.

Another way speech recognition can get tripped up is in working with a caller or agent with an unfamiliar accent. The way we speak makes us all unique, but for AI transcription services, it adds a layer of difficulty. Have you ever used a “machine voice” in order for robotic agent to understand you over the phone? For recorded calls, accents can pose a problem. Many transcription systems support numerous accents, but you need to specify what it is before you start the analysis. This isn’t always easy to automate and will continue to be a challenge until speech analytics become sophisticated enough to determine one from another.

Voxbone's Speech Analytics platform is integrated with leading providers

More devious ways to confuse a machine-based transcription service may also arise, as a pair of Facebook AI researchers found. The study found that certain manipulations undetectable to the human eye can “trick” algorithms into ignoring or misinterpreting data from photographs–and they can be adapted for speech recognition. All it takes is a layer of digital noise to knock a functioning transcription service off the rails.

Since call quality is key for accurate speech recognition, and CRM integrations are becoming more vital each day, an important consideration should be made for the kind of phone systems a company uses. VoIP offers a compelling alternative to traditional, PSTN telephony, at less cost.

In the hands of consumers, it allows calls to be made from smartphones and even connected devices like the Amazon Echo. On the business side, using VoIP in a contact center is becoming the best option for interconnected systems that can be highly automated and provides the foundation for both recorded conversations to be transcribed later, or real-time analytics that offers agents immediate feedback or guidance for each call.

Keep an ear to the ground

Despite AI’s shortfalls and even its potential to “be fooled,” the costs more than justify the results. As equipment costs go out the window, replaced by cloud-based services that enable recording and analytics, the possibilities for growth and improvement are monumental. Customer support and customer retention go hand in hand, so the more valuable insights into improving the experience, the better.

Sure, the phone is old fashioned when you think of it as a clunky piece of hardware with limited capabilities. But as a modern business tool, and the current channel of choice for customer service agents and salespeople, voice interactions are a treasure chest of useful data–and the right speech analytics service holds the key.