ellipse
pattern
Nov 04, 2020

Self-Serve Customer SatisFAQtion Starts with Natural Language Understanding

Reginald Tang, Senior Product Manager, AI

Self-Serve Customer SatisFAQtion Starts with Natural Language Understanding

Data is not information, information is not knowledge, knowledge is not understanding, understanding is not wisdom.

Aside from sounding like a passage from the Tao Te Ching, the above passage—usually attributed to astronomer and systems analyst Clifford Stoll—does a good job of succinctly describing a ladder of value in resolving customer support issues, especially in self-service channels. Customers want actionable responses to their natural language questions, not reams of search results or an endless FAQ page.

Another quote, this one definitely from Clifford Stoll, drives home the above point and lays out the criteria for a truly satisfying customer support experience:

“Information, unlike data, is useful. While there’s a gulf between data and information, there’s a wide ocean between information and knowledge. Knowledge—not information—implies understanding. And beyond knowledge lies what we should be seeking: wisdom.” [Emphasis mine.]

Wisdom is probably a long way off for AI. But what technologies are at hand that can layer understanding onto mere information—and thus provide actionable knowledge on demand?

Enter Natural Language Understanding (NLU)

NLU technology, whether used by a chatbot or an intelligent search box, makes it possible to take a customer's query (the "utterance") and figure out its intended purpose (the "intent"). 

For example:
Customer utterance: "How do I get  there?"
System response: "Our address is 114 Main Street." (System displays map and link to directions.)

Such goal-oriented interactions hinge on first figuring out intent. An NLU-equipped system doesn't require customers to explicitly navigate menus or issue specific commands such as “NEED LOCATION.”

By contrast, a keyword-based (non-NLU) system would probably come up empty (since the question lacks unique words or phrases) or, worse, might return thousands of matched results ... results a customer would consider useless noise.  Rest assured, if a customer did type "NEED LOCATION" into a chat or search box, NLU would respond with a single, helpful street address.

What to Look for in NLU-based Customer Support Solutions

Because NLU is so valuable compared to static FAQs or standard website searches, the market for NLU-based customer support solutions is very competitive. Herewith some guidance.

Consider systems that supply the following capabilities …

  • Omnichannel: Works in every channel from a web chatbot to Facebook Messenger or Apple Business Chat—even when typed into a search box.
  • Conversational: Enables journeys across multiple turns—back-and-forth exchanges with a virtual or live agent—and guides the customer to a successful conclusion.
  • Ambiguity-averse—Clarifies your meaning by asking questions, such as “It sounds like you could be asking about [X or Y]. Which one do you need help with?” Of course nothing is perfect, and sometimes a user’s utterance can pertain to more than one intent.

 

The Secret to Customer SatisFAQtion? An NLU-Enhanced Knowledgebase

NLU enables your use of another transformative technology: The knowledgebase, or KB.

A KB adds structure and classification to assets—such as articles, support tickets, forums, and documents—for easy search and retrieval. Using NLU to expose this impressive collection of institutional know-how is a game changer for customer support. Why?

  • This virtual agent drastically lowers support costs compared to live chat agents or phone support. An NLU chatbot or search box decides customer intent by asking plain-spoken questions.
  • Knowledge management tools serve up high-quality, up-to-date content to the virtual agent.

The above combination enables virtual agents to supply the appropriate, actionable information customers need to achieve their goals.

Such a system is valuable only to the extent it easily and quickly combines knowledge and understanding and deploys it to the right customer channels. Given how fast information changes in large enterprises, deployment should be measured in days, not months.

Therefore, when shopping for a KB system to power a customer support solution, be sure to look for the following features …

  • Per-channel automatic content adaptation: Choose a KB that lets you author content once and then displays it correctly in each channel (according to that channel’s conversation design guidelines). As messaging apps proliferate, and vendors push more content and interactions into chat bubbles, each channel displays response content to customers differently.
  • Flexible structure with shared content: Choose a KB whose structure is able to align seamlessly with your existing organization and with processes for curating and securing information. Ensure that content common to all structures updates once and shows up everywhere.
  • Rapid and continuous deployment: A KB needs robust publishing capabilities to ensure customer-facing virtual agents serve up the latest customer FAQs the moment the content is finalized—with no waiting period for IT or the solution provider. Quite literally, every second counts when servicing thousands of customer interactions per day.

 

Take the Next Step

[24]7 Answers, a foundational [24]7.ai Engagement Cloud product, accelerates how you create and share knowledge to support your customers.

For more [24]7 Answers information, visit the website, download the data sheet, or write to us at [email protected].

Related Posts

Personality is Not Optional in Chatbot Design
Nov 26, 2020 Personality is Not Optional in Chatbot Design

“No response is a response. And it’…

What to Bot—and Not
Oct 23, 2020 What to Bot—and Not

What if I told you that, with the…