<?xml version="1.0" encoding="UTF-8" ?>
<!-- Created from PDF via Acrobat SaveAsXML -->
<!-- Mapping Table version: 28-February-2003 -->
<TaggedPDF-doc>
<?xpacket begin='﻿' id='W5M0MpCehiHzreSzNTczkc9d'?>
<?xpacket begin="﻿" id="W5M0MpCehiHzreSzNTczkc9d"?>
<x:xmpmeta xmlns:x="adobe:ns:meta/" x:xmptk="Adobe XMP Core 9.1-c001 79.675d0f7, 2023/06/11-19:21:16        ">
   <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
      <rdf:Description rdf:about=""
            xmlns:xmp="http://ns.adobe.com/xap/1.0/"
            xmlns:xmpMM="http://ns.adobe.com/xap/1.0/mm/"
            xmlns:stRef="http://ns.adobe.com/xap/1.0/sType/ResourceRef#"
            xmlns:stEvt="http://ns.adobe.com/xap/1.0/sType/ResourceEvent#"
            xmlns:dc="http://purl.org/dc/elements/1.1/"
            xmlns:pdf="http://ns.adobe.com/pdf/1.3/">
         <xmp:CreateDate>2024-01-12T14:24:44+05:30</xmp:CreateDate>
         <xmp:MetadataDate>2024-01-12T14:24:46+05:30</xmp:MetadataDate>
         <xmp:ModifyDate>2024-01-12T14:24:46+05:30</xmp:ModifyDate>
         <xmp:CreatorTool>Adobe InDesign 19.1 (Macintosh)</xmp:CreatorTool>
         <xmpMM:InstanceID>uuid:a72a1cb1-b95a-41f7-ac09-b92de9cb6fb7</xmpMM:InstanceID>
         <xmpMM:OriginalDocumentID>xmp.did:e2c787e8-e602-4bc7-bd33-4b0641be08c2</xmpMM:OriginalDocumentID>
         <xmpMM:DocumentID>xmp.id:35b4b329-f0ea-4782-b545-f1aa20f3b578</xmpMM:DocumentID>
         <xmpMM:RenditionClass>proof:pdf</xmpMM:RenditionClass>
         <xmpMM:DerivedFrom rdf:parseType="Resource">
            <stRef:instanceID>xmp.iid:913d148f-6619-4b22-bf38-9e0ceb0e9e42</stRef:instanceID>
            <stRef:documentID>xmp.did:30593138-2aa7-4032-a386-3eaada4cd092</stRef:documentID>
            <stRef:originalDocumentID>xmp.did:e2c787e8-e602-4bc7-bd33-4b0641be08c2</stRef:originalDocumentID>
            <stRef:renditionClass>default</stRef:renditionClass>
         </xmpMM:DerivedFrom>
         <xmpMM:History>
            <rdf:Seq>
               <rdf:li rdf:parseType="Resource">
                  <stEvt:action>converted</stEvt:action>
                  <stEvt:parameters>from application/x-indesign to application/pdf</stEvt:parameters>
                  <stEvt:softwareAgent>Adobe InDesign 19.1 (Macintosh)</stEvt:softwareAgent>
                  <stEvt:changed>/</stEvt:changed>
                  <stEvt:when>2024-01-12T14:24:44+05:30</stEvt:when>
               </rdf:li>
            </rdf:Seq>
         </xmpMM:History>
         <dc:format>xml</dc:format>
         <pdf:Producer>Adobe PDF Library 17.0</pdf:Producer>
         <pdf:Trapped>False</pdf:Trapped>
      </rdf:Description>
   </rdf:RDF>
</x:xmpmeta>
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                                                                                                    
                           
<?xpacket end="w"?>
<?xpacket end='r'?>

<P>WISE INNOVATION DOI: 10.59571/mpi.v1i2.5 </P>

<P>Automate empathy: </P>

<Part>
<H1>Five steps to turn online customer rage into rave </H1>
<Figure>

<ImageData src="images/MPI I2E5_img_0.jpg"/>
</Figure>

<Sect>
<Sect>
<H3>R. Raghuram &amp; Janakiraman Moorthy </H3>
</Sect>

<Sect>
<Sect>
<H4>Problem of practice: </H4>

<P>The 
<Link>customer rage survey 2023</Link>
 reports 74% of customers experienced problems with products or services last year, more than double of the figure in 1976.1  Even more alarming is the statistic that 63% of customers who experienced a product failure reported feeling rage, which leads to an increase in aggressive behavior during problem resolution. Dissatisfied customers are not silent sufferers – 79% make an effort to voice their complaints. The impact of vocal customer dissatisfaction has only multiplied, thanks to the deep penetration of smartphones and social media platforms. In our essay, we highlight a possible solution – derived from an 
<Link>innovative study</Link>
 by Dennis Herhausen and team.2 The insight suggests that a prompt, empathetic response that reflects the client’s emotional state can go a long way in placating the irate customer. </P>
</Sect>

<P>1 Herhausen, D., Grewal, L., Cummings, K.H., Roggeveen, A.L., Villarroel Ordenes, F. and Grewal, D., 2023. Complaint De-Escalation Strategies on Social Media. Journal of Marketing, 87(2), pp.210-231. </P>
</Sect>

<Sect>
<Sect>
<H3>Online Dissatisfaction </H3>

<P>For companies, dealing with customer dissatisfaction is just the first hurdle. The customer survey cited above also revealed that 69% of dissatisfied customers were not content with monetary compensation alone; they sought more empathy from the company. When combined with the high rates of those dissatisfied customers who want to give voice to their issues, it points to a troubling phenomenon: an unhappy customer no longer wishes to interact with emotionless, mechanistic AI-driven Chatbots. Instead, they yearn for genuine and empathetic guidance. </P>

<P>Since the survey was first conducted about fifty years ago, technology has amplified the impact of dissatisfied customers. Fifty years ago, an unhappy customer would share their grievances with around 10 friends and family members. A decade ago, with the rise of social media, socially connected customers reached an average of 280 individuals with each comment or post. Today, thanks to smartphones and the instantaneous nature of platforms like Twitter, a customer's complaint can go viral within minutes. Compared to a decade ago, when only 5% of customers were online, today that number has skyrocketed to 50%. And a third of dissatisfied customers do take to social media to publicly express the problems they faced. </P>

<P>Consequently, the damages a company suffers due to grievances voiced by dissatisfied customer now have a global reach.  In September 2016, Delta Airlines, suffered a 
<Link>5-hour computer outage</Link>
 that caused thousands of flight </P>

<P>cancellations and delays, 
<Link>costing the company $150 million </Link>
in lost business and refunds.3 Passengers were left stranded at airports, and the company's customer service channels were overwhelmed. Angry customers took to social media to express their frustration, which quickly spread, leading to massive social media criticism.4 Delta was not alone: In the US alone, firms face a potential loss in business value of $887 billion due to improper complaint handling post a service or product failure. </P>

<P>And the problem is global: In 2019, OYO, an Indian hospitality company, faced severe backlash when a customer shared images and videos of an 
<Link>unsanitary and </Link>

<Link>unhygienic hotel room</Link>
 booked through the platform.5 The images went viral on social media, leading to 
<Link>widespread </Link>

<Link>criticism of OYO</Link>
's quality assurance and customer service practices.6 </P>

<P>Such instances of businesses facing losses due to customers’ ire poses critical questions for businesses— Do managers have effective mechanisms to manage such a crisis? In 2023, just the opposite happened with global confectionary brand 
<Link>Cadbury’s handling</Link>
 of an influencer’s video that called out the brand on its ‘health drink’.7 Cadbury’s response of denial created more social media controversy than the original post itself. This is in stark contrast to how 
<Link>Cadbury responded</Link>
 to another serious issue with its Dairy Milk chocolates controversy twenty years ago.8  Hence there is a need to address issues such as — Do decision-makers possess the necessary tools to predict and address customer outrage based on online posts? Is it possible to pacify customers quickly in such situations to tamp down the damage? </P>

<P>The research by Herhausen's  team provides an innovative solution to all of the above challenges. They propose a Natural Language Processing (NLP) algorithm that not only helps to pacify customers but also kindles a sense of gratitude among those who have been wronged. The way companies can overcome the negative signal is to practice active online listening and organization-wide empathetic responses to consumers. This innovation enables CMOs to design appropriate customer response systems to effectively handle social media complaints. </P>
</Sect>

<Sect>
<H3>Handling Customer Rage with NLP </H3>

<P>Social media conversations are now vital part of modern communication, but they differ significantly from faceto-face conversations. Understanding these differences is crucial for businesses to maintain a positive reputation and enhance customer satisfaction. Unlike face-toface conversations, social media conversations are text based, which creates unique challenges such as enhanced public visibility but time lag in responses. Furthermore, such negative complaints expressed due to high-arousal emotions like anger, disgust, or anxiety can significantly impact the purchase intentions of other customers. </P>

<P>Hence, here is a five-step process companies can follow to ensure their response to negative customer feedback tamps down customer dissatisfaction. </P>

<P>Step 1: </P>

<P>Respond promptly to negative social media posts. Delaying a response increases the likelihood of other customers joining the conversation, potentially amplifying the negative sentiment. India’s leading telecom provider, 
<Link>Airtel</Link>
, received multiple customer service complaints, including a recorded conversation with a client and a customer service representative who used offensive language during the interaction.9 The audio clip went viral, and posters joined in to share their opinion on the company’s customer service practices. The company could have de-escalated the social media flak, if it had been prompt in handling the issue. </P>

<P>Step 2: </P>

<P>Detect complaints based on expressed emotions on social media. NLP algorithms, when combined with specialized dictionaries, can help companies identify and categorize complaints as high-arousal emotions (i.e. anger, disgust) or low-arousal emotions (i.e. sadness, disappointment). </P>

<P>
<Link>Several banks and Indian e-commerce giant Flipkart </Link>
</P>

<P>have used NLP to generate insights and identify potential problems, as well as their severity.10 </P>

<Sect>
<H5>Step 3: </H5>

<P>Avoid monetary compensation as a first response.While monetary compensation is effective in pacifying angry customers face-to-face, it is a costly solution and may invite fraudulent claims. As the rage survey revealed, often such a response is not the best one – As a more powerful approach is empathetic guidance, using the steps that follow.  </P>
</Sect>

<Sect>
<H5>Step 4: </H5>

<P>Active listening can play a significant role in placating enraged customers. When customers feel that a company is genuinely paying attention to their concerns, their extreme emotions subside. Active listening implies asking clarifying questions, paraphrasing, and adapting the language of response to the customer.  While responding to social media complaints, companies can improve the active listening response by using the NLP algorithms (as depicted in Fig 1) to respond in a style —formality vs. informality, complexity vs. simplicity, and concreteness vs. abstractness— that matches the complainant’s. Here, companies can signal to the customer that they are actively listening to their complaints. </P>
</Sect>

<Sect>
<H5>Step 5: </H5>

<P>Expressing empathy is the most important tool in pacifying angry customers. Understanding the customer’s emotions is more effective than explaining service failures. Companies can develop or utilize an existing empathy word dictionary and incorporate these words in their reply. Once they have crafted a response that aligns with the appropriate style and includes empathetic language, it can be shared with the complainant. Empathy has a more significant impact on increasing the likelihood of de-escalating consumer emotions than active listening. A mere 1% increase in empathy can lead to a staggering 90% increase in the probability of customer gratitude, while a 1% increase in active listening yields a 14% increase. </P>
</Sect>
</Sect>
</Sect>

<Sect>
<Sect>
<H3>When and How to Use NLP </H3>

<P>Active listening and empathy play a significant role in reducing the negative emotions of complaining customers. However, their impact varies depending on the type of </P>
</Sect>

<Sect>
<Sect>
<H5>Figure 1: Process of Detecting and Mitigating Customer Rage with NLP </H5>

<Sect>
<H5>Customer Complaint </H5>
<Figure>

<ImageData src="images/MPI I2E5_img_1.jpg"/>
</Figure>
</Sect>

<Sect>
<H5>NLP ALgorithm </H5>

<P>Emotional words dictionary </P>
<Figure>

<ImageData src="images/MPI I2E5_img_2.jpg"/>
</Figure>
<Figure>

<ImageData src="images/MPI I2E5_img_3.jpg"/>
</Figure>

<P>Emotion </P>

<Sect>
<H5>Classification </H5>

<P>Complaint Style </P>
<Figure>

<ImageData src="images/MPI I2E5_img_4.jpg"/>
</Figure>
<Figure>

<ImageData src="images/MPI I2E5_img_5.jpg"/>
</Figure>
</Sect>
</Sect>
</Sect>

<Sect>
<Sect>
<H5>Firm Response </H5>
<Figure>

<ImageData src="images/MPI I2E5_img_6.jpg"/>
</Figure>

<P>Source: Created by the authors on the basis of Herhausen, D., Grewal, L., Cummings, K.H., Roggeveen, A.L., Villarroel Ordenes, F. and Grewal, D., 2023. Complaint De-Escalation Strategies on Social Media. Journal of Marketing, 87(2), pp.210-231.  </P>

<P>emotions expressed, whether high-arousal emotions like anger or low-arousal emotions like disappointment. Understanding this distinction is crucial for companies aiming to address customer concerns and earn gratitude effectively. A company should recognize that different emotional states require tailored approaches. </P>

<P>Similarly, when it comes to active listening, only a high level of active listening, with a Linguistic Style Matching (LSM) score above 0.85, increases the probability of earning an angry customer's gratitude. However, for disappointed customers, a medium level of active listening increases the odds of earning their gratitude. However, when active listening does not occur, negative customer emotions can snowball. Take the 
<Link>case of Ryan Block</Link>
, a customer who tried to cancel his subscription with Comcast, a cable TV and home internet service company. The customer service representative did all but cancel Ryan’s subscription.11 The </P>

<P>High negative arousal </P>

<P>Product/ Service Failure Low negative arousal Online complaint </P>

<P>Offline/</P>

<P>No complaint </P>

<P>Offline/</P>

<P>No complaint </P>

<P>Online complaint </P>

<P>NLP based focused intervention </P>

<P>Traditional intervention </P>
</Sect>

<P>Face to face/ </P>

<P>Difficult to detect</P>

<Sect>
<P>NLP based focused intervention </P>

<P>Traditional intervention Satisfaction increases </P>

<P>Either </P>

<P>satisfied or unsatisfied</P>

<P>Satisfaction plateaus and decreases </P>

<P>Either </P>

<P>satisfied or unsatisfied</P>

<P>call was recorded and uploaded online, where it went viral and further damaged Comcast's already poor customer service reputation. </P>

<P>In conclusion, tailoring responses to customer emotions is paramount in effectively handling complaints. Recognizing the differences between high-arousal and low-arousal emotions allows companies to customize their replies, utilizing appropriate empathy and active listening. By doing so, companies can increase their chances of pacifying customers, resolving their concerns, and earning their gratitude. </P>
</Sect>
</Sect>
</Sect>
</Sect>

<Sect>
<Sect>
<H3>Challenges in Deployment </H3>

<P>An important challenge in deploying the proposed NLP based solution, is developing a specialized dictionary that the algorithm can draw from. Each product market has its own lingua franca or jargon, and specific customer pain points, which necessitate the customization of dictionaries to include words and phrases specific to that particular industry. By tailoring the language resources to the industry, companies can better understand the unique context of customer complaints and provide more satisfactory responses. </P>
</Sect>

<P>A related implementation challenge is for global companies, or for those that operate in multiple languages. Even a single country can be multilingual, for instance in a country like India, implementing NLP solutions has recognized the development of such 
<Link>dictionaries in local languages</Link>
.12 The proposed solution involves creating multiple dictionaries for negative emotions and empathy and identifying word structures to determine the complaint style accurately. Developing and validating dictionaries in local languages will empower companies to effectively address complaints in a manner that resonates with customers, leading to better satisfaction and brand loyalty. </P>
<Figure>

<ImageData src="images/MPI I2E5_img_7.jpg"/>
</Figure>

<Sect>
<P>Cultural nuances also play a significant role in the effectiveness of empathy. The level and expression of empathy can vary across cultures, with some cultures valuing expressed politeness as a norm. In such cultures, even a moderate level of empathy from a company can effectively pacify customers. Understanding and adapting to these cultural differences is crucial for companies operating in diverse markets to ensure their responses align with customer expectations and foster positive customer experiences. </P>

<P>Raghuram R is an Assistant Professor of Marketing at SPJIMR. </P>
</Sect>

<P>Janakiraman Moorthy is a Professor of Marketing and Department Chair  at SPJIMR. </P>

<P>This article may contain links to third party content, which we do not warrant, endorse, or assume liability for. The author’s views are personal. If you have some inputs you would like to share, you can also reach out to us mpi@spjimr.org </P>

<Sect>
<Sect>
<H5>References </H5>

<P>1 W. P. Carey School of Business at Arizona State University, “Historic National Customer Rage Survey: Record Level of Product and Service Problems Incite Surly Customers to Yell More and Seek Revenge for Their Hassles,” March 7, 2023, https://www.prnewswire.com/news-releases/ historic-national-customer-rage-survey-record-level-of-product-and-service-problems-incite-surly-customers-to-yell-more-and-seek-revenge-fortheir-hassles-301765053.html. </P>

<P>2 Dennis Herhausen, Lauren Grewal, Krista Hill Cummings, Anne L. Roggeveen, Francisco Villarroel Ordenes, and Dhruv Grewal, Complaint De-Escalation Strategies on Social Media (August 1, 2022). </P>

<P>3 Chris Isidore, “Delta: 5-Hour Computer Outage Cost Us $150 Million,” September 7, 2016, https://money.cnn.com/2016/09/07/technology/delta-computer-outage-cost. </P>

<P>4 Phil Helsel and Jay Blackman, “Delta Computer Outage Results in Nationwide Ground Stop,” September 26, 2018, https://www.nbcnews. com/storyline/airplane-mode/delta-computer-outage-results-nationwide-ground-stop-n913151. </P>

<P>5 Ranganath Guptha Ragavendra Singri, “Oyo - Unhygienic and Filthy Room,” 2018, https://voxya.com/view-complaint/unhygienic-and-filthy-room/27548/. </P>

<P>6 Shubham Bhatia, “Poor Customer Service, Cancelled Bookings: What Ails OYO Rooms in India?,” Newslaundry, October 28, 2019, https://www. newslaundry.com/2019/10/28/from-poor-customer-service-to-cancelledbookings-what-ails-oyo-rooms-in-india. </P>

<P>7 Tanzila Shaikh, “Bournvita Gets Bitter over Sugar Controversy: What Should Have Been the Right Response? - Exchange4media,” April 20, 2023, https://www.exchange4media.com/marketing-news/bournvita-gets-bitterover-sugar-controversy-what-was-the-right-approach-126808.html#:~:text=As%20expected%2C%20the%20brand%20responded,legal%20notice%20 from%20the%20brand. </P>

<P>8 Manoj, “How Cadbury Pulled Itself Out of Its Downfall,” Blogs - Vedak (blog), March 27, 2020, https://vedak.com/blogs/cadbury-pulled-itself-outof-its-downfall/. </P>

<P>9 Pranali Lotlikar Chindarkar, “Fine of Rs 25K Levied on Airtel for Threatening, Mentally Harassing Consumer,” March 22, 2018, https://www.dnaindia. com/business/report-fine-of-rs-25k-levied-on-airtel-for-threatening-mentally-harassing-consumer-2258683. </P>

<P>10 Abhijit Ahaskar, “Companies Use NLP-Based Sentiment Analysis to Source Intelligence,” Mint, August 2, 2022, sec. Technology, https://www. livemint.com/technology/tech-news/companies-use-nlp-based-sentiment-analysis-to-source-intelligence-11659419241250.html. </P>

<P>11 Steve Rose, “Comcastic: The Excruciating Customer Services Call That Went Viral,” The Guardian, July 17, 2014, sec. Technology, https://www. theguardian.com/technology/shortcuts/2014/jul/17/comcast-customer-services-call-ryan-block. </P>

<P>12 Uma Kannan, “Embracing Regional Language Tech for Wider Reach- The New Indian Express,” July 7, 2023, https://www.newindianexpress. com/lifestyle/tech/2023/jul/07/embracing-regional-language-tech-for-wider-reach-2592236.html. </P>
</Sect>

<Sect>
<H5>Article Information: </H5>

<P>Date article submitted: July 4, 2023 Date article approved: September 25, 2023 Date article published:  October 10, 2023 </P>
</Sect>
</Sect>
</Sect>
</Sect>
</Part>
</TaggedPDF-doc>
