For many contact center (CCaaS) product teams, ASR is little more than a 'hygiene' capability they need to secure quickly and efficiently so they can continue to build. This approach is about to become costly, as artificial intelligence (AI) innovations, combined with ASR, look set to supercharge the sector with the advent of Speech Intelligence.

An almost limitless market

Retaining customers is far cheaper than converting them, and customer satisfaction and retention is a key metric for many brands; which means maximizing the customer experience across multiple touchpoints.

This can be a tall order for contact centers, which are often inundated with requests. Most customers are happy to wait about three minutes for an operator, so an average call center hold time of 13 minutes makes for a poor customer experience – and the potential for some not-very-nice interactions when agents do become available.

As such, CCaaS platforms that can support agents to deliver exceptional service quickly and efficiently, while maximizing the experience of the agents themselves, are valuable resources and priority investments for brands the world over.

AI adoption in call centers

Artificial intelligence (AI) has proven to be something of a game-changer for CCaaS. AI chatbots and virtual assistants now routinely help to contain calls, providing timely, tailored support without the need for a human agent. This improves call resolution rates and protects brand reputation.

Agents are a key component of the customer experience, and the widespread adoption of AI has facilitated the development of tools to support them, too. AI routing, which identifies customer issues and ensures they're put through to the right agent, reduces customer frustration and means agents can resolve problems more efficiently, improving their working environment. AI is also removing much of the administrative burden associated with call center operations, automating compliance requirements, analytics, and reporting.

Automatic speech recognition (ASR) is the technology that underpins each of these applications. Capable of transcribing speech-to-text, ASR is a critical component of CCaaS platforms. But not all ASR is built equal and, while CCaaS product teams understand the importance of the functionality it supports for their end users, a race to the bottom on cost has devalued the technology itself.

Rethinking ASR

The spoken word is our primary means of communication, each of us has our own style of speech that reflects our personalities, backgrounds, and lived experiences. To successfully make use of the spoken word, it needs to be captured accurately and fully understood. Contact centers serve a diverse cross-section of customers, and CCaaS solutions must be able to reliably capture and interpret all varieties of speech.

The importance of CCaaS platforms for brands from almost every industry makes for a highly competitive marketplace, and end-product pricing is reflective of that. While product teams have dedicated resources to building features that differentiate them in the market, the procurement priority has been sourcing ASR as cost-effectively as possible. So much so that it's often viewed as little more than a transcription service.

But ASR is critical.

Models with low accuracy invariably result in faults with features downstream. At best, this becomes an annoyance experienced by some agents and customers. At worst, it prejudices users against the feature and platform entirely – who is going to waste their time with a tool that routinely fails to work for them?

Speech Intelligence and CCaaS

For CCaaS teams that want to differentiate their product, provide real value to customers and increase their total addressable market, the solution is to invest in speech technology – not skimp on it. Speech is our core means of communication and there is a wealth of insight available in audio data. Combining ASR with AI innovations, including large language models, is how CCaaS product teams achieve this, harnessing speech data and transforming it into a uniquely valuable asset for their customers.

We call this Speech Intelligence. It uses techniques, particularly AI, to deliver value from speech data, building capabilities such as interpretation, translation, automation and pattern recognition on top of transcription to support new and unique features.

CCaaS product leaders can leverage Speech Intelligence to build valuable, differentiated features with speech. For example, analytics can be broadened to include multiple transcripts, rather than just assessing one at a time. This gives end users the ability to pull combined insight from every call over a specific time period. It also means product teams can build speech capabilities on top of transcribed audio from multiple calls, leveraging functionality like sentiment analysis to provide insight into customer attitudes overall, or showcase the sentiment when certain topics are mentioned, such as products, pricing and delivery.

As a group, call center agents have a shockingly high attrition rate, which now tops 33% in the US and 24% in the UK. The result of low morale and a highly competitive job market puts pressure on call center operators, increasing training and recruitment costs, reducing productivity, and negatively impacting customer experience.

Speech technologies have the potential to mitigate this. Supervisors can monitor agent performance and provide tailored support. Sentiment analysis of calls, for example, can be used to understand whether certain teams are exposed to more negativity than others and rotate staff accordingly. It can also be used to identify training opportunities, flagging if an agent turned a conversation from a negative sentiment to a positive one so others can analyze and learn how they did it.

Similarly, functionality like summarization – automatically creating a short overview of a conversation – can be used to reduce the administrative burden on agents, negating the need to manually write up notes and ensuring a consistent customer experience.

Real-time transcription not only allows for quick summarization after calls, but can also support innovative features that end users won't be able to get anywhere else. This might include a tool that recognizes keywords within a customer's speech and can search for associated internal content, giving agents instant access to the information they need to respond to a customer's request. Or, it might help to support agents who are working in a second or third language, providing real-time translations to their native language so they can clearly understand technical requests, avoid miscommunications, and provide the best possible customer service.

What can you achieve with Speech Intelligence?

ASR can deliver more than just transcription, it is the foundation for Speech Intelligence. This collection of technologies can deliver value across an almost limitless number of use cases, providing CCaaS product leaders with the tools to meaningfully differentiate their products and platforms.

With highly accurate ASR, Speech Intelligence can become a competitive differentiator, effortlessly opening up international markets and creating a platform end users simply can't live without.