GenAI - 3) Gen AI chatbot case studies - Different levels of efficiency

 3) Gen AI chatbot case studies - Different levels of efficiency

 With continuation to our last blog on pain points and tips to use chatbots in support teams, we continue here with classic examples which you’d have come across when chatting with one. 


To introduce the baseline, support teams work closely with the following teams. There might be more, but we focus here on the core roles.


1) Workflow management and operations teams to balance contact volume and manpower requirements

2) Content Management Team (CMS) for regularly updating content as per new updates and feedback from quality assurance teams

3) AI engineers who work on the bot’s functionalities and improvisations.


Here, we have 4 case studies by deep diving on training models from few reputed brands. This article analyses how the bot handles the contact and the resulting user experience.


Case 1 - User X contacts the support team to know updates about a specific policy. To his dismay, the bot typed the entire answer in the chat box which is quite long and difficult to read from a mobile device.


Case 2-  This time, the bot gave the link of the concerned policy page in the chat box directly.


Case 3 - The chatbot had no free textbox, customer has to select the issue from the available options. The loop ends here, if the user need live support he has to call.


Case 4 - The chatbot had an option called ‘My issue is not listed’, and offers a textbox for typing the user's concerns. If the answers are not available and the user specifically types words like ‘callback’, ‘live support ’, the chat is connected to a CSA with notes written in chat.


Write down your results, put yourself into the customer’s shoes and speculate on the following factors as you do


  1. Scope for result accuracy

  2. Personalized results

  3. Feedback credibility

  4. Right support


Here are few cases where support teams deploy bots to simplify and automate operations 


  1. Data retrieval (or) RAG (Search info for help questions)

  2. Text summarization ( Summarize an article into simple chat)

  3. Auto - route contacts (Used in GCCs to optimize workforce)

  4. Automate communication


Before we deep dive into the solutions for the above case studies, here are few clues to derive your answers on how the user experience would be


  1. Customer obsessed companies value the time and effort of their customers irrespective of the reason of contact - They try to keep it minimal for the user

  2. The bots are trained in NLP (Semantics and Sentiment) specialization modules to decide when to transfer a chat to live support

  3. They present their replies in a very crisp manner, beyond just simply fetching it from their knowledge base.


Well, I guess these would have thrown some light onto the case studies. Let’s see what would have been done better in the above scenarios.


Case 1 - Implementation of design thinking in UI and optimizing it. The text should have been summarized than copy pasting


Case 2 - RAG has been implemented here which saves time in searching. However, it should have been summarized and presented in a chat box directly to save the effort of reading a long article.


Case 3 - Provide a free text box always to hear the customer’s issue in his own words. This adds authenticity to the feedback, and will be very helpful for doing VoC (Voice of Customer) analysis. Understand the user issue and connect to live support if needed


Case 4 - Mark the user issue code, record the user feedback typed in his own words and connect to live support if needed.


Despite the tips mentioned above, it is always advisable to audit the responses manually, at least in a random sampling method even if not fully. Also, it's crucial to keep upgrading the AI capabilities (For instance, enhancing sentiment analysis and semantics) based on the user feedback.


To summarize, using AI to simplify the regular workflow brings efficiency in operational level, but curating the user experience is what truly brings the customer delight factor for support teams.


Said that, are there any golden rules to be followed for handcrafting the best virtual assistants ?


How do we identify use cases which can be disruptive inventions ?



Stay tuned !! 

Comments

Post a Comment

Popular posts from this blog

GenAI - 2) Identifying problem statements in using virtual assistants / support chatbots:

  2) Identifying problem statements in using virtual assistants / support chatbots: In the last edition, we saw an example where User X orders food from a reputed platform, and he tries to reach out to support for help. With the options the bot offers, there is no solution and the loop continues. (Bot asks to call the delivery agent where the user reached out as he is unable to).. Giants like Amazon invest significant efforts to continuously audit and improve their customer service everyday - to be identified as earth’s number one customer centric company. But why is customer service considered so important in building a brand / company’s identity ? Whether it’s a manufacturing / service sector, it’s very possible that there are going to be errors - in the process or quality of offerings to their customers. Collecting feedback from users is crucial to introspect, iterate and improve the process and stay ahead of competitors and retain, expand the customer base. Inbound feedback is ...
Read more

Gen AI - 1) How cool are Gen AI based customer service chatbots ?

  How cool are Gen AI based customer service chatbots ? With the bloom of Gen AI based automated solutions in almost every sector, the transition from live support to automated chatbots in customer service / technical support areas is phenomenal over the past few years. Players with automated chatbot services like Freshdesk, Zendesk etc have a solid role to play in reshaping the user experience for new age users, who value time and effort more than money. But, are we able to provide the best to them, let’s check it out.          Before we deep dive into the subject, here are some interesting findings about chatbots in customer support. The global chatbot market size was estimated at USD 5,132.8 million in 2022 and is expected to expand at a compound annual growth rate (CAGR) of 23.3% from 2023 to 2030 59% of customers expect a chatbot to respond within 5 seconds  62% of consumers would rather talk to a chatbot than wait 15 minute...
Read more