COMPLAINING ABOUT? UNDERSTANDING THE REASONS FOR CUSTOMER - PowerPoint PPT Presentation

WHAT CUSTOMERS ARE COMPLAINING ABOUT? UNDERSTANDING THE REASONS FOR CUSTOMER DISSATISFACTION. Dr. Gennadi Lembersky Bar-Ilan, 2014

NICE, The Worldwide Leader of Intent-Based Solutions NICE ENTERPRISE Customer Interactions NICE SECURITY Safety & Security Driven by x-channel and multi-sensor NICE analytics, we enable organizations: ACTIMIZE  Improve business performance Financial Risk &  Increase operational efficiency Compliance  Prevent financial crime  Ensure compliance  Enhance safety and security 2

>150 150 >80% >80% ~25,00 25,000 0 Countries Customers Fortune 100 Customers ~3,40 3,400 0 $1B $1B NASDAQ: Employees 2013 Non-GAAP Revenues >10 >1000 00 >35 Service Experts Local Offices 3

Collect Cross-Channel Customer Data Billing IVR Transactions Phone Calls Interactions Emails Chat Customer Engagement Sessions Analytics CRM Customer Data Branch Web Visits Transactions Mobile

Customer Feedback  Short messages sent by SMS or e-mails or entered via a special form on the website.  “Price an service. I hav ben with XXXXX since u opend up here 1518 years ago an hav njoyd ur changes an improvments ”  “I think its horrible that XXXXX charges a 3.00 fee 2 allow u 2 PAY ur bill the XXXXX locations..ridiculous.u have 2 pay 2 pay”  Typos, abbreviation, phonetic writing  Wrong\missing punctuation, no spaces between words  Lower\Upper case confuses POS taggers 5

Agent Notes  Agent summarizes the interaction with the customer either during the phone call or immediately after it.  XXXXXX customer phone is giving him prob educated customer that we will credit half of 354$$$ Credit for note 3// to customer can Jump into new device will go in store// educated cust// explained hw it would wrk// ive opts// closed 6

Agent Notes  In how many ways one can misspell the word payment: paymetn paymnt paymnet pyment paymet pyament payent payemtn paymen payament pymtn payt pament payment pymt pytm paymnent apayment apayment payment pa 7

Automatically Transcribed Phone Calls  my name is i don't to better sister  my name is aida to better assist you  i'm trying to get count as an assistant status  i'm showing that the account is in a suspended status 8

Text Analytics 9

Sentiment Analysis 10

USE CASE: WHAT CUSTOMERS ARE COMPLAINING ABOUT? 11

Understanding the reasons for customer dissatisfaction  Consider thousands of customer interactions  Assume we can extract relevant text fragments from them  Consider linguistic variability and domain specific details

Customer Complaints Word Cloud 13

Closer Look  Food is really poor . Lets the rest of the service down .  Service was good but food was awful .  It's conviniant and runs on time . Seating and food can be improved  Food could be a bit better in standard leisure . I like travelling by Quasigo . You should have more connections into europe . Would choose you over fllying ... 14

Excitement: Using entailment graphs  Let us consider two fragments “ Meal needs to be improved in Standard Magnum ”  “ Better food at Stars Bridge station ”   They clearly talk about the food quality problem, with different details Meal needs to be improved in Standard Magnum Better food at Stars Bridge station

Excitement: Using entailment graphs  Let us consider two customer interactions  “ Meal needs to be improved in Standard Magnum ”  “ Better food at Stars Bridge station ”  The corresponding entailment graph is produced Meal needs to be Better food at improved in Stars Bridge Standard station Magnum Better food = Meal needs to be improved

Customer complaints application 17

THANKS! 18