Conversational analytics is not a new technology. Contact centers have used it for over 15+ years.
In doing so, they have analyzed customer conversations – across all engagement channels - digging up precious insights.
For example, they have determined: why are people contacting us? Why has traffic spiked? What’s making customers unhappy? The list goes on.
The catch is that these conversational analytics systems have historically hinged on expansive natural language processing (NLP) models.
Consequently, their development required expensive, specialist staff. Also, they offered limited language support and narrow vertical applications.
Now, those limitations are practically gone with generative AI (GenAI).
The Brakes Are Off
Traditionally, running a contact center required on-prem virtual and bare metal servers. Adopting advanced analytics under that model was prohibitively expensive.
Now that everything is cloud-based, implementing many AI features is as simple as ticking a box, according to Carl Townley-Taylor, Product Manager at Enghouse Interactive.
“There’s no need for additional configuration or expensive on-prem infrastructure,” he said.
“Additionally, contact centers can test out conversational analytics solutions without committing to massive upfront investments.” - BLOCK
Then, there’s GenAI. While it hasn’t changed the user interface (UI) of conversational analytics systems all that much, it’s pushing their functionality forwards rapidly.
Indeed, vendors no longer need extensive research and development (R&D) to support new languages or domains. They can get those up and running quickly because large language models (LLMs) handle that for them.
Consequently, those providers can focus on ensuring data sovereignty, creating effective visualizations, and bringing use cases to customers faster.
Auto-Categorization: A Game-Changing New Feature
As noted, creating NLP models for conversational analytics systems once involved extensive manual configuration. That made industry-specific solutions a rarity.
After all, developers had to understand the nuances of each industry, like product names or customer types. Yet, LLMs have enabled auto-categorization, which is proving a game changer.
Sharing an example of how this works, Townley-Taylor said: “We [Enghouse Interactive] partnered with an Australian cattle company. They wanted to categorize interactions based on product names - different types of cattle, in this case.
“Previously, this would’ve required building custom models to account for all the variations and nuances,” he continued. “With large language models, we simply provide parameters, and the system generates results.”
“Deployment is faster, costs are lower, and the same technology can easily be adapted for other industries.” - BLOCK
With this capability, the technology will also drive deeper insights. After all, if a business can categorize interactions by product name, sentiment, or time of day, it can identify new performance trends.
3 Conversational Analytics Use Cases to Get Started With
Now that conversational analytics systems have become more accessible and industry-specific, more contact centers can start to test it out.
