Slightly further along in the journey are companies that are still using old school methods for assessing contact center operations . They ’ re using manual QA to evaluate a random sampling of calls per agent per month and are having their agents select the reason for calls , resulting in limited analysis and insights .
At the middle of the maturity curve are companies using surveys to collect customer insights . This is an important start , but the results usually reflect the experience of a minority of customers , typically highly satisfied or dissatisfied customers .
At the top of the contact center maturity curve are the most advanced companies : those that are monitoring and analyzing 100 % of the conversations their customers are having with their team across touchpoints , including emails , support tickets , live chat , phone calls , SMS , and social channels .
These businesses are able to do this at scale by using AI and ML models . These tools minimize the need for humans to manually read and tag every conversation . They instantly transcribe and draw insights from phone conversations , video , text , and both structured data ( such as drop-down choices in a menu ) and unstructured data ( blocks of open-ended text , such as that in a social media mention ).
KEY USE CASES FOR AI AND ML IN THE CONTACT CENTER
Tools like AI and ML can support several key use cases in contact center operations .
1 . REDUCING TIME TO UNLOCK
INSIGHTS AND VALUE These technologies work faster and are more scalable than requiring humans to manually review conversations . AI can make it possible to automatically assess every contact center interaction and instantaneously categorize the reason for contact and sentiment of conversation .
Being able to understand the context of all customer interactions enables organizations to save time on completing manual and time-consuming tasks , understand customer needs faster , and implement changes to improve CXs in the moment .
2 . UNCOVERING EMERGING OR PRE- VIOUSLY UNKNOWN ISSUES NOT YET ON A COMPANY ’ S RADAR
38 CONTACT CENTER PIPELINE
At the start of the COVID-19 pandemic , companies weren ’ t necessarily prepared for the kinds of questions customers would have around safety precautions , such as cleaning protocols or mask requirements .
For companies with AI tools in place to analyze contact center conversations , it quickly became apparent that these were the types of issues customers were concerned about .
But as the economy shifts , and other factors influence the concerns of today ’ s customers , businesses using AI can catch wind of these kinds of trends in the moment and react more nimbly than competitors without such timely insights .
3 . DETECTING CUSTOMER CONVERSATIONS THAT MIGHT OTHERWISE BE OVERLOOKED Companies have their own way of referring to their products , services , and employees , but customers don ’ t necessarily use the same terminology . They might use acronyms , abbreviations , misspellings , or different words to refer to the same things .
When companies initially set up search queries to analyze their contact center conversations , these typically focus on the words and phrases the company uses . AI can be used to find the correlation between different concepts and words to make sure companies are capturing all customer conversations related to a specific topic , even when consumers use new or different terms .
4 . CREATING CONNECTIONS WITH
CUSTOMERS Better connecting with customers is important for customer service . When customers feel a connection to a company , they feel a higher sense of loyalty .
As discussed earlier , AI can be used to automatically analyze the emotion of any conversation , and companies can use these insights to detect behavioral patterns across segments .
This knowledge empowers them to understand how people of different age groups , stages of life , regions , races , and genders feel when interacting with the contact center . They can then adapt agent training accordingly so they can better handle certain situations , especially as sensitive topics arise .
WHERE DOES CHATGPT FIT INTO TODAY ’ S CX ?
ChatGPT is one of the latest technologies that ’ s taken the internet by storm , and amidst all the buzz there have been bold claims about how the tool is going to replace everyone from doctors to engineers .
The reality is , AI needs to be guided by humans , especially in the contact center . No form of AI will replace the need for contact center employees .
Large language models ( LLMs ) like ChatGPT are only as good as the data sets they ’ re modeled off of . ChatGPT in particular is open source , which means it ’ s not a secure environment . Therefore , no sensitive customer data should be shared with the platform ( or any other open-source platforms like it ) for analysis purposes .
In the future , LLMs that do meet privacy and security standards for sharing and analyzing customer data may be useful for the contact center . For instance , these technologies might be able to better answer questions that have historically been difficult to address consistently and reliably without bias or uncertainty , such as calculating first call resolution ( FCR ) more accurately at scale .
Today , companies rely on imperfect methods for evaluating FCR . This is done by asking agents to self-report whether a conversation has been successfully resolved upon the first contact , surveying customers about whether their issue has been resolved on the first attempt , or making an educated guess based on whether customers call back again or not .
A LLM might be able to understand the full context of conversations and detect whether a question or issue that ’ s raised gets resolved , based on the back and forth between the agent and the customer .
Some brands are already starting to use LLMs as the primary method for evaluating conversations and assessing whether there has been successful FCRs . This is based on analyzing the transcripts of the conversations as well as any additional contextual data , such as tickets that have been created in association with the interactions .