Control Tower
In today's digital landscape, delivering exceptional customer service requires more than just phone lines. Modern contact centers leverage the power of AI-powered applications to streamline operations and elevate interactions. These applications empower agents with real-time insights, personalize service for each customer, and automate repetitive tasks. This results in faster resolution times, increased agent productivity, and ultimately, happier and more satisfied customers.
Challenge
Building an enhanced version of Contact Center application using Gen AI to ease the day to day tasks of the customer success representatives.
Current Problem
The Contact center representatives are attending around 30-40 customers daily and during peak season this rises upto 50+ a day and to keep a track of every call manually in itself become daunting.
Their strain to manage high volume of frustrated customers, identifying the most suitable solutions on a case-by-case basis, and manually tracking every interaction can be overwhelming.
Besides, answering routing enquiries, composing emails, tickets, and planning the next appropriate steps leads to unnecessary consumption of valuable bandwidth as well as can cause potential errors and mismatched solutions almost common for all customers eventually slowing down entire operation.
This is not healthy for companies especially in this decade when customers expect invest in experience rather than product especially to get personalized interactions. As per a report, around 76% of customers get frustrated when it doesn’t happen.
Solution - What and How to improve?
In order to streamline and automate this system, utilizing Gen AI is most viable solution as of today. It can revolutionize contact centers by automating ticket creation and instantly providing relevant solutions to agents. This will not only ensure consistent service for all customers, but also frees up valuable agent time for more complex issues.
The agents won’t need to jot down points discussed over call, AI will generate real-time transcripts and summarise essential points out of it basically capturing crucial call details.
This will let the AI craft accurate and timely responses, draft ticket bodies, and even propose email content, all based on the conversation.
The Agents will only have to approve or edit it.
When it comes to follow up on emails, a time threshold can be fed to AI post that it can directly trigger a follow up email to the respective team.
For routine inquiries about tracking, payments, or refunds, AI can directly retrieve the information and deliver the appropriate status and solutions to customers, eliminating the need for agent intervention.
For call scripts, agents can simply provide keywords, and AI will tailor call scripts based on each customer's unique interaction pattern.
It can also suggest insights and offer by analysing the CSAT scores and purchase pattern of the customers
Smart Search: Instead of searching for information on the portal, an AI chatbot can do that for the agent, and provide with solutions on the go in the same workspace. Also, it can analyze recent live chats to suggest helpful keywords, helping the agents to find relevant content they need faster.
For team meetings, AI can automatically generate agendas and recommend participants based on past meeting history.
Almost half the users that contact a brand on social media expect a response in under 60 minutes and out of these, 80% expect a reply within the same day. Integrating a multi-channel support can help meet this expectation
By analyzing customer interactions, it can identify areas for improvement and generate reports for relevant teams.
Design Iterations
Presented the above findings to the Leadership team and came up a with below wireframes after multiple rounds of discussions and brainstorming.
Impact
By building a GenAI integrated system, most of the non-complex tasks are taken care by AI so a significant amount manual effort should be reduced resulting in improved following parameters.
Improved FCR (First Call Resolution): A higher FCR rate with efficient problem-solving and reduces the need for follow-up calls.
Reduced Average handle time (AHT): The AHT will be should at least match the ideal 6 min standard.
Maintained Occupancy rate: The improved bandwidth utilization should match the MO rate which is supposed to be at least ~80%
In a nutshell, above improved parameters would also expedite the support quality and elevate the response time, providing extraordinary customer experience.
As per a survey report 58% of US customers are ready to pay more to a brand if they can provide a better customer experience
Integration
Tools and techniques: Platforms like Rasa Stack (open-source) or Amazon Lex (cloud-based) will be required for building the right chatbot which can be trained to understand customer queries by identifying relevant keywords using natural language processing (NLP) techniques.
Data Requirement: This can be done by feeding historical customer interaction data like chats, emails, call recordings or transcripts for training the LLM and chatbot.
This will help the AI to understand the type of assignments the Agents are handling currently.
When a keyword is detected, the chatbot can trigger the large language model (LLM) API to fetch information or relevant knowledge base articles or FAQs for the agents to share with customers.
Future scope
A common issue faced by Agents is transferring calls to cross departs when it is dialled to wrong team so before routing the call to Agents, AI can check for the issue and either can resolve it directly if its a routine enquiry or route to the right team rather than teams routing it to each other internally.
Moreover, almost 62% of millenials and 75% of Gen-Z customers prefer self-service almost all the time, even when they have an option of contacting support so this feature is definitely an add to improvise customer service level.
However, the ASA(Average speed of answer) which is typically around 28 seconds or less in the industry should be as minimum as possible as it enhances customer satisfaction and reduces abandonment rates( should be less than 5%)
