Speech analytics isn't completely new, but an integration with machine-learning technology has enabled Marchex to build out a platform that enables marketers to visually analyze customer sentiment and a variety of interactions on phone calls to businesses.
Marchex Speech Analytics aims to help marketers optimize media spend by seeing a visual representation of the purchase funnel. It integrates with the company's Omnichannel Analytics Cloud launched earlier this year.
The main purpose of the Marchex Speech Analytics platform is to analyze a variety of interactions such as lost opportunities, calls where the customer failed to complete the transaction, or high-intent calls to determine traffic sources driving sales.
Jason Flaks, senior director of speech analytics at Marchex, calls the analysis "training data" for chatbots and machine-learning technology, in addition to its intended use, which is to help brands optimize their media spend and better understand how campaigns perform through the feedback of consumers who interact with the company via phone calls.
Despite the platform's more traditional use, Flaks acknowledges that the data analysis Marchex provides can be useful for marketers that are thinking through how they would build a chatbot for apps or Web sites. Chatbots and artificial intelligence technology all require training data for the ways that humans interact with each other to provide the best customer experience.
"If I'm bidding on a keyword on Google and it doesn't look like it's performing well, you need to know how it's performing based on the performance of the keyword; or maybe the keyword performs fine, but the store locations do not answer phone calls coming in," Flaks said. "This way you know there's no issue on the keyword not performing, but rather the store clerk not answering the call."
Caleb Williams, marketing innovation manager at Two Men and a Truck, uses Marchex Speech Analytics -- specifically the Lost Opportunities Dashboard feature -- to identify what areas of its marketing campaigns work,
The data also provides insight into how consumers react to promotions and products.
Built with audio signal processing analysis, the technology does not analyze inflection points in words, but it does determine when phrases are positive or negative. The company does intent to inlay in the future other audio-based signals, per Flaks.
The technology isn't equipped to analyze voice search on smartphones or home hub devices like Google Home or Amazon Echo. It really focuses on transcribing conversational speech for marketing- and sales-driven phone calls in a call center during a business to consumer interaction.