In the midst of the customer care chaos that accompanied the global pandemic, Interactions LLC launched an automated Virtual Collections Assistant, VCA for short. Much like the Guest Experience Platform (GXP), which Interactions introduced last year, the VCA is purpose-built to meet the well-understood, but growing, challenges of a specific industry’s use cases.

In the case of restaurants, they were able to increase phone based sales by eliminating unanswered calls, shortening the time on hold and reducing resulting call abandonments. Once engaged in a conversation with the customer, the IVA uses the combination of machine learning and human assistance to make recommendations, ensure the accuracy of input and accelerate the payment process.

Better than the Best Agent At Scale

In the case of the VCA, the primary problem for companies in the Accounts Receivable Management (ARM) domain has involved detection and elimination of calls that put live agents in touch with “the wrong parties.” But that task could be performed by a robotic dialer with strict rules in place to authenticate the recipient of a collection call. By contrast, the VCA is a human-like virtual agent designed to be “better than the best live agent” consistently and at scale.

In the most challenging days of pandemic-induced activity, Interaction’s go-to-market partner and customer is ERC (Enhanced Resource Centers) , a 20 year-old business process outsourcer that has experienced great success in the Collections and Accounts Receivable Management domain. Marty Sarim, the president and CEO of ERC, considers Interactions’ VCA an “exponential technology.” After many months of operations, ERC discovered that their VCA, called EVA, was more than up to the task of detecting “wrong parties” and correctly routing calls. It was, and is, able to understand the intent of legitimate callers and handle a variety of common functions, including customer authentication, payment processing and a growing number of tasks involved with handling fraud, disputes, deaths and bankruptcies.

His company reported an eye-popping 60% reduction in connecting agents to “wrong numbers”. Most importantly, he noted that “most customers would rather interact with a virtual agent immediately than wait in a queue or gat a call back later from a live agent.”

Measurable Results

As evidence of EVA’s efficacy, a case study from ERC listed the following measurable improvements:

• 27% drop in inbound calls directed to agents: which is the result of key task completion by VCA
• 30% drop in the probability of a caller talking to the wrong agent on the first attempt
• 15% increase in the “agent conversion rate”, meaning that the company is recovering more revenue

Deployment of VCA by ERC moves an intelligent virtual assistant to the front lines of customer contact in a domain where inbound traffic, via phone, text and messaging has moved to a high level and the mix of calls has altered significantly. It serves as a reality check for firms that have historically used the term “chatbot” to describe their intelligent assistants (IAs) or intelligent virtual agents (IVAs). Managers have had low expectations for chatbots and customers, too often, see them benignly as a novelty and less kindly as a nuisance.

For both debtors and collectors, VCA is a welcome intermediary. It is a non-judgmental agent programmed to understand and respond to the terms that customers are comfortable using. At the same time, it spares getting agents involved, except for the sorts of corner cases that require human intervention. Interactions takes an approach that, by definition, is human-assisted artificial intelligence. It is proving to be very well-suited to the challenges of surrounding ARM.

Categories: Intelligent Assistants, Intelligent Authentication