Chatbots in Customer Service

Topic: 
|
2 min read
Philipp Heltewig
Media contact: Philipp Heltewig July 19, 2018

SEVERAL LEVELS OF SOPHISTICATION ALLOW FOR LOW-FRICTION EARLY ADOPTION

Chatbots have seen tremendous interest over the past years, with several applications emerging as low hanging fruit for organizations with customer-facing operations.


Such operations are usually rely heavily on human agents who handle customer service requests via phone, email and chat. This approach is effective for the customer but comes with several significant drawbacks, such as agent availability and slower resolution times even for trivial requests.

Chatbots or conversational AI technology in general can help overcome these issues by augmenting existing, human-based customer service operations with AI-based, automated technology. Such technology can be applied in different stages as organizations mature in their adoption of conversational interfaces.

 

Woman connecting cellphone and laptop computer  

 

  1.  FAQ Chatbot

    A common first step in chatbot applications is the exposure of a company’s existing Frequently Asked Questions (FAQs) through a conversational interface, typically as a chatbot on a website. The bot can classify the customer request against a set of predefined FAQs and then provide the correct answer. AI is used to find the correct FAQ from the customer input.

    Set-up time: days
    Sophistication: low

  2. Transactional Chatbot with and without human handover

    Transactional bots help to answer more complex customer service requests, such as resetting a password, finding a shipment or changing an order. Such capabilities require the integration of existing backend systems, such as databases, APIs, ERPs, CRMs and the like, allowing the bot to read and write data to line of business systems. Sometimes these systems are integrated with human-to-human chat systems to provide help from a human agent if the customer gets stuck.

    Set-up time: weeks
    Sophistication: medium-high

  3. Fully integrated conversational AI solution

    Fully integrated systems can take transactional and other requests on multiple channels, such as chat, smart speakers and phone systems (IVR). The integration with phone systems is particularly interesting, as many organizations already have call centers in place and can reduce load on their human agents by handling many requests in an automated fashion.

    Set-up time: weeks to months
    Sophistication: high
The presented stages are not exclusive, but instead build on each other. An organization might start out with an FAQ bot, then extend it to handle transactional requests and later integrate the AI solution into all of their support channels.