Invoca Inc. today extended its call tracking and analytics platform with capabilities that apply keyword analytics to transcribed conversations and automatically categorize them for marketing purposes.

The new Signal Discovery feature uses unsupervised machine learning to scour large volumes of recorded calls and automatically group them based upon words and speech patterns.

Unsupervised machine learning discovers patterns without explicit instructions from a human operator. The data derived from these analytics can then be used to refine customer experience and marketing programs.

Invoca is targeting the roughly 40 million calls monthly that it says are driven by search results and tools as such as Google LLC’s “Click-to-Call“ feature. Those calls are often recorded, but defining useful marketing metrics from them is challenging because transcription is expensive and the content is unstructured.

Invoca’s technology analyzes a database of previous conversations to create clusters of topics based upon speech patterns (pictured). Marketers can use these clusters to detect conversational events like callers asking about price matching or competitive promotions.

“This creates a conversational map of clusters of unique topics,” said Ian Dailey, the company’s senior director of product marketing. “Marketers can choose to use this historical information to tag inbound calls.”

Invoca estimates that marketers can boost conversion rates 15% to 20% by using speech analytics to enhance campaigns. Coincident with today’s announcement, the company is releasing the results of a study of 500 business-to-consumer marketers that found that only 8% say phone call data is their top source in informing campaign personalization.

The benefits of informed call routing also pay off in operational efficiency by enabling calls to be directed to the employees best equipped to handle them. “By understanding what types of calls you’re driving you can stop having salespeople field calls meant for customer service,” Dailey said.

He cited the example of one hospital that had full-time staff members spending hours each week listening to call recordings to determine whether paid media campaigns were resulting in booked appointments. Invoca’s software  determined that many callers hung up when their calls were routed to a messaging service. The organization changed its routing rules and dramatically improved the ratio of calls to appointments booked.

The technology takes advantage of targeted phone numbers and advertising features that route calls based upon the webpage upon which the call originated. “If you know that every time someone calls in from a certain landing page you can deflect to the sources that are best prepared to answer those questions,” said Sean Storlie, senior director of product management.

Each conversation is tied to a digital customer ID, making it possible for marketers to create or expand customer relationships based upon information stored in customer relationship management tools and social media interactions. For example, a customer who asks about competitive promotions can be targeted with a personalized advertisement or webpage about that topic.

Founded in 2008, Invoca has raised $60 million and says it processes about 100 million calls annually. Its cloud-based service is priced by a combination of features ordered and inbound call volume, but the company didn’t provide further details.

Image: Invoca