How Artificial Intelligence Is Used In Customer Experience Automation

Artificial intelligence and virtual agents promise improved user experiences and decreased servicing costs. These occur through several pathways.

First, virtual agents can be used to ensure the customer is routed to the proper department. Virtual agents are conversational computer programs that interact directly with a customer without human intervention. They are also known as “front end bots”, “virtual assistants”, or “automated assistants”. Many call centers have separate channels for sales and support, routing the customer incorrectly harms both the business and consumer. An incorrect routing frustrates the customer, increases the transaction time, and requires the engagement of multiple agents that adds to costs. Utilizing a virtual agent to pre-screen calls ensures that users are directed properly.

Virtual agents also allow for up-scaling with minimal cost adjustments. This is particularly useful during unexpected times of heavy call traffic. Virtual agents working in tandem with humans can manage spikes in demand without the need to hire additional staff. Relatedly, virtual agents reduce the overall number of employees needed by independently managing high-volume, low-value transactions. This frees companies from the hiring, training, and assimilating of new employees and yields significant cost savings.

Finally, virtual agents have been shown to benefit the contact center employees as well as their customers. Associates are no longer faced with managing the same trivial queries day-in and day-out and are less likely to develop “associate fatigue”. The result is higher quality conversations when a transaction is escalated to a live agent.

Evolution From Human To Bot

Customer service has traditionally been a field that relies on the empathic abilities of a human being to resolve issues. Yet as the costs for human staffed call-centers have grown, automation has become imperative and led to the introduction of bots. A bot is a software application that runs automated tasks over the Internet at a much higher rate than would be possible for a human. Today’s bots are becoming increasingly complex and functional.

There are two primary models businesses use when incorporating automation into their customer service departments: the “bot-only” model and the “bot-assisted agent” or “cyborg” model. Both have pros and cons and selecting the correct option depends upon the overarching goals of the business.

In the bot-only model, a conversational computer program interacts directly with a customer without human intervention. A “bot-assisted agent” is a human agent supported by bot technology. Other terms for the model include “cyborg”, “human in the loop”, and “hybrid model.” The bot advises the agent on the best course of action or automates knowledge search and other agent functions. At present, the consensus is that bot-assistant hybrids are more effective than bots acting independently, but only if there is seamless transitioning between bot and human agent. LivePerson reported a 30 – 35% increase in efficiency using this model.

The biggest challenge for purely virtual bots is processing natural language and interpreting customers’ questions. Clients and businesses often do not use the same language when describing a problem. As a result, they see a very high error rate. This challenge may be alleviated in the future as increasingly sophisticated artificial intelligence bots with stronger natural language processing (NLP) and natural language understanding (NLU) are developed.

Types Of Automation In Customer Service

AI-driven customer service can be further broken down into six categories of automation.

1. Multiple choice chatbots

Though frequently referred to as “AI” technology, these manually scripted bots contain no intelligence. Instead, they are used to simplify transactions by using a more conversational form of input and output. The retail, publishing and travel industries make frequent use of this type of bot.

2. Virtual agents

Virtual agents (VAs) have been on the market for several years, but they remain essentially glorified search engines able to parse FAQs. The language programming of most VAs is designed to match keywords rather than infer meaning from a request.

3. Bot-assisted agents

These “human in the loop” or “cyborg” models are the next step up from virtual agents. This symbiotic relationship allows bots to increase the efficiency of conversation while a human agent monitors quality control and gives customers both correct and personal answers.

4. Real-time emotional analysis

Emotional analysis software interprets customer emotions, intentions and social signals in real time enabling increased customer and agent satisfaction. This technology is still early, but reported results from companies adopting the software are promising.

5. Intelligent assistants

As contrasted with virtual assistants, intelligent assistants are the ultimate end-goal and comprise a fully-automated solution. Improvements in natural language processing (NLP) will advance the market to this eventually, current technology typically have unacceptably high error rates.

6. Content curation

Content curation is distinct from the five already listed. Primarily useful to online retailers, a content-curation bot would learn based on a customer’s answers to questions and/or previous purchases in order to offer personalized product suggestions.