One of the most compelling use cases for enterprise bots is customer support.
Traditionally, as companies grow in revenue, customer service costs grow in tandem.
Companies constantly seek ways to minimize these servicing costs through nearshoring, offshoring, and now customer service bots.
Companies ranging from RBS to KLM and Disney Stores to Overstock are eagerly innovating with intelligent bot technologies, hailed as the solution to scaling customer care delivery.
There are two basic types of customer service bots: “front end bots” and “bot-assisted agents”.
A “front end bot” is a conversational computer program that interacts directly with a customer without human intervention.
They’re also known as “virtual assistants” or “automated assistants”.
A “bot-assisted agent” human agent is supported by bot technology. Other terms for the model include “cyborg” or “human in the loop.”
Over the last decade, countless websites have implemented virtual agents or “front end bots.”
7 is a leading customer service company that has created more than 180 chatbots for companies such as Duke Energy, CIBC, RBC, and the Disney Store.
Daniel Hong, the Senior Director of Product Marketing Strategy at 7, explains, “These automated website chat agents handle the first level of queries such as FAQs.
When the Virtual Agent doesn’t know the answer, it transfers the user to a real human agent.”
Using chatbots to automate answers to basic customer questions decreases the average Agent Handle Time (AHT) by 10% or more.
Despite the advances in NLP and AI, many of these front end bots cannot fully understand complicated user inquiries.
A customer contacts support because they have an issue, and an unintelligent bot that fails to answer the question leads to more frustration.
Therefore, some companies implement “bot-assisted agents” instead.
Robert LoCascio, the CEO of LivePerson, notes that “Customer satisfaction on the traditional front-end bots is below 70%, which is really low for customer care and sales.
We think the best way is a hybrid model called cyborg, which is having the bot and the agent working in tandem next to each other.”
In this cyborg model, the bot interprets the conversation and suggests choices to the human agent.
The bot can also change the reply format based on the inquiry platform, like elaborate longer in an email and keep Twitter responses to 140 characters.
The human agents determine whether the answer fits the question and add further elaboration if needed.
Rather than search their knowledge base for an answer and generate a custom response each time, agents simply monitor the bot’s answer.
The artificial intelligence learns from the agent’s customizations to incrementally improve the automated answers.