you will apply the concepts, methods, and tools studied in this course to produce a PowerPoint presentation and with speaker notes. The primary objective of this Project is to convey that you understand Natural Language Process, the steps in the marketing analytics process, and best practices in data visualization.
The analytical approach you discovered this week is natural language processing (NLP.) With this process in mind, you will develop a comprehensive PowerPoint presentation related to your findings.
Your PowerPoint presentation should include the following:
An introduction to your favorite large retail organization, including the size of the company, its revenue, and its ‘owned’ digital and social media channels. Identify if the organization uses cognitive analysis, which was covered in Module 4.
Collect and present at least eight (8) reviews from your retailer related to one product from the company website. You can also use Yelp, Google, and social media channels. Include a listing of your reviews on your slide.
Following the concepts in chapter 10, assess and clean the text data from the reviews. Incorporate concepts used, including, Tokenization, Stemming, Lemmatization, and N-grams.
Describe the terms that appeared most frequently.
Describe the stop-words that you removed.
Identify the term(s) that appeared most often.
Review the ideas presented in the textbook for a term-document matrix (TDM.) There are three necessary steps: (1) tokenize, (2) create vocabulary, and (3) match and count.
Develop your TDM by adding your findings from the documents examined. (An Excel template is available for
Your TDM would look similar to the one in Exhibit 10-5 from the textbook.
Copy and paste your completed TDM from Excel directly to your slide. Or you can Save As a JPEG to use in your presentation.
Analyze the findings of your cleaned data.
Suggestions: Describe how your findings might be helpful to your retailer. (You will pitch your findings to senior leadership, which will include relevant content from slides 7, 8, 9, and 10.)
Identify what data source(s) would be most useful based on your suggestions. (For example: customer service transcripts, user-generated content, social media comments, etc.)
Describe what insights would be needed to address the issue. (For example, adding a new feature to an existing product.)
Based on your analysis, describe how customers feel about the product.
Identify what future implications and growth opportunities are present.
Create a data visualization, and be creative! (Login to your Tableau account used in the hands-on experience in Module 5, Excel, or other tools.) Paste your data visualization directly to your slide.
Key takeaways from the NLP process.
PowerPoint presentation (PPT, PPTX, PDF)
Term-document matrix file
Data visualization file
*** The book is ***
Essentials of Marketing Analytics
byJoseph F. Hair
TDM template DOCUMENTSSTOP WORDSTERM -DOCUMENT MATRIX Doc 1Stop Words RemovedTermsDoc 1Doc 2Doc 3Doc 4Doc 5Doc 6Doc 7Doc 8 Doc 2 Doc 3 Doc 4 Doc 5 Doc 6 Doc 7 Doc 8 10.1What Is Natural Language Processing? With the rise of the internet and social media, a large amount of data is increasingly available for marketing decision making. About 25 percent of the data is structured and therefore directly usable by most statistical analysis methods. Recall from Chapter 1 that structured data is made up of records that are organized in rows and columns. This type of data can be stored in a database or spreadsheet format so it is easy to analyze. The format of structured data is numbers, dates, and text strings that are stored in a clearly defined rows and columns. In contrast, unstructured data, which represents more than 75 percent of the emerging data, does not have a predefined structure and does not fit well into a table format (within rows and columns). Moreover, unstructured data requires further processing before it can be analyzed using statistical methods and when possible must be converted into structured data prior to analysis. This type of data includes, for example, text (words and phrases), images, videos, and sensor data. Customers, employees, and competitors produce a large amount of unstructured data that can be very useful in marketing decision making. Unstructured data is, therefore, increasingly being used to develop more effective marketing strategies. The focus of this chapter is on using natural language processing to identify patterns in text data. Social media posts, product reviews, emails, customer service call records, sales calls, and chatbots are just some common sources of text data. The sources shown in Exhibit 10-1 produce a huge volume of text data useful for solving marketing problems. Exhibit 10-1 Example of Text Data Sources How can insights be obtained from text data? Natural language processing (NLP) is a branch of artificial intelligence (AI) used to identify patterns by reading and understanding meaning from human language. Through NLP, companies can analyze and organize internal data sources, such as customer service calls, as well as external data sources, like social media content. Analyzing and interpreting text data provides answers to a variety of questions and produces insights about the company, customers, products, services, and market. NLP can be used by marketers to extract information that can be applied in both proactive and reactive ways, including responding to customer comments and complaints and building on product preferences. Ultimately, NLP can help companies reduce potential conflicts and prioritize the level of urgency when responding to concerns and opportunities. 188Page 337 Consider these examples: • Brand reputation: What is customer sentiment or emotion toward the brand? How much of the conversation is focused on the brand versus other issues? • Customer satisfaction: How do customers feel about their in-store or online experience? • Market intelligence: What are competitors doing and how are customers responding? Are certain topics or keywords trending? • Product or service quality: Do customers prefer certain product features or are there quality issues? PRACTITIONER CORNER Jasmine Jones | Database Administrator at MetLife Jasmine Jones Jasmine Jones graduated from Elon University in 2019 with a bachelor’s degree in computer science. While there, she nourished her love for data by minoring in data science and most importantly, working at the university’s Center for Analytics. During that time, she was able to advance her data analytics skills by working with companies, both large and small, doing tasks such as coding, text mining, and web scraping in order to advise their decisions with the use of data. She is now fortunate to continue working with data in her first career after college as a database administrator at MetLife, further nurturing her love for all things data. Many companies are implementing NLP-driven interactions with customers. From chatbots to conversational analytics platforms used to derive sentiment analysis and monitor social media chatter, companies are using NLP to understand customers, create efficiencies, and personalize service. How can natural language processing be applied to help companies engage and interact with customers? Natural language processing enables companies to engage and interact with the customer in a pointed manner. The best way to engage a customer can be determined by understanding the interests of the customer with tools such as topic modeling and sentiment analysis. Once engaged, companies are able to have more meaningful interactions with their customers with the help of chatbots that have the customers’ best interest in mind and continue to learn and adapt with each customer they interact with. How do marketers use NLP to create a successful customer journey quickly, and with minimal friction? Marketers utilize natural language processing to better understand their customers. Marketers can analyze a large number of product or service reviews without much effort. They can do this by utilizing tools such as sentiment analysis, which will allow them to quickly understand whether a product was perceived positively or negatively, or tools such as topic modeling to suggest other products related to their interests. What is the potential for natural language processing in marketing? Natural language processing can help marketers extract insights quickly from social media posts, online forums, chatbot conversations, call center transcripts, emails, and so on. These insights can reveal what customers like or dislike about the company. Such understanding can be used for advertising, marketing campaigns and to personalize the customer’s experience. Page 338 Is there a unique use case that you could share? Marketers can generate new advertisements using information from natural language processing tools such as topic extraction and summarization of customer conversations. They can use what the customer cares about to create new advertisements. The customer reaction to the new advertisements can be captured and analyzed in real-time, which makes it possible to test the efficacy of new advertisements quickly and effectively. 10.2How Is Natural Language Processing Used in Practice? Marketers use NLP to understand the “what,” “how,” “when,” and “why” of customer and market behavior. For example, marketers can use NLP to extract underlying intentions, beliefs, and attitudes behind the behavior. Whether companies are using chatbots or monitoring social media, companies can analyze text data records to learn more about what their customers want and need. NLP has been applied in a variety of contexts, as described in the following text. Optimize Inventory and Engage Customers in Marketing Campaigns Do you ever mention your cold or flu on social media? Clorox uses social listening to monitor discussions occurring on social media that might impact its brands.8 Applying NLP, Clorox analyzes social media conversations around symptoms that could accompany these viruses, including most recently with the COVID-19 pandemic. By combining this information with data from the Centers for Disease Control and Prevention, the company assists retailers in understanding appropriate inventory levels and targeting certain customers with marketing campaigns for products such as Clorox disinfectant wipes. Produce New Products to Meet Customer Needs The hashtag #sriRancha began trending on Twitter, which captured the popularity of mixing ranch dressing and hot sauce. Social media chatter indicated customers were creating their own flavors of Hidden Valley’s ranch dressing (see Exhibit 10-2). The company listened to the conversation and seized on the opportunity to develop new products such as siracha-flavored ranch dressing.9 Exhibit 10-2 #sriRancha on Twitter Source: Kate @ChemistK8, Twitter, April 22, 2015, https://twitter.com/ChemistK8/status/590786488826273793/photo/1. Page 339 Simplify Guest Travel to Improve Hospitality Chatbots, driven by NLP, have become common for the hospitality industry. Companies like Marriott are using chatbot virtual assistants to enhance connections with customers and answer questions any time of the day. These chatbots can understand and respond in different languages. They also recognize certain behaviors and understand preferences to then make recommendations. The goal of chatbots is to engage customers when and where they want and to personalize service. 189Create a Better Experience for Customers The success of Uber is highly dependent upon the reliability and ease of use of the technology that powers their mobile phone application. Uber gauges driver and user experiences using Brand 24, a social media monitoring technology. Following the rollout of a new app, Uber observed a spike in people talking about the company on social media. Social media mentions went from around 9,000 instances to almost 22,000 within the first few weeks after the app release.10 But the insights for Uber were not limited to the increase in number of mentions. They also obtained both positive and negative sentiments, enabling them to improve the app and be more responsive to customer needs. Thus, Uber was able to determine whether the app created a better customer experience or whether they needed to make additional improvements. Add Unique Features to Products Companies often monitor social media to see what customers are saying about their products. How do customers rate a product compared to competitive products? What features do they like or want, or what do they dislike? Ford uses NLP to make discoveries in social media conversations.11 When Ford decides to add new features to products, they often monitor conversations about a similar option on another vehicle. For example, adding features such as power lift gates or seat heaters can be costly if customers are not receptive to the change. Understanding what customers really want helps Ford determine if the change is worth pursuing. Improve Customer Service Companies such as Avis and Dish Network use chatbots to enhance customer service. The use of chatbots powered by NLP enables companies to respond to customers quickly, manage a large volume of simple inquiries, and reduce operating costs. If a customer has questions outside of business hours, the chatbot is available. In addition, chatbots enable customer service representatives to prioritize responses and stay focused on addressing more complex customer issues. For example, Dish Network’s chatbot manages about 30 percent of the incoming volume, whereas Avis chatbots can manage almost 70 percent of customers’ questions.12 When a higher percentage of inquiries are addressed by chatbots, companies can reduce costs and satisfy more customers. Facilitate Customer Ordering “Alexa, can you order a pizza?” “Alexa, can you turn off the lights?” Companies such as Amazon and Domino’s Pizza use voice recognition software so customers can order and track products. Domino’s uses technology powered by natural language processing to facilitate customer orders for pizza using voice commands.13 Customers can easily customize, select pickup locations, and pay using their natural language app. Domino’s and other companies have now extended their ordering capability to virtual assistants such as Amazon Alexa (see Exhibit 10-3) and Google Home. This method has been so successful, the company started testing similar voice-guided technology for incoming phone orders. Of course, when deciding to implement NLP, companies must consider whether a particular customer segment is receptive to that approach of responding, because some segments prefer traditional person-to-person communications. 190Page 340 Exhibit 10-3 Ordering Domino’s Pizza through Amazon Alexa Source: “Domino’s Anywhere,” Domino’s Pizza, https://anyware.dominos.com. Strengthen Customer Relationships The Royal Bank of Scotland uses text data from various structured and unstructured data sources when applying NLP.14 Customer emails, surveys, and call center interactions can be examined to discover reasons behind customer dissatisfaction. By effectively uncovering the drivers of customer dissatisfaction using NLP, the company is able to understand service failures, resolve issues faster and with greater success, and improve profitability. 10.3How Is Text Analytics Applied? Text analytics is an NLP technique that includes text acquisition and aggregation; text data preprocessing through cleaning, categorizing, and coding; text exploration; and text modeling (see Exhibit 10-4). Page 341 Exhibit 10-4 Text Analytics Steps Step 1: Text Acquisition and Aggregation The first step in text analytics is to determine the marketing analyst’s business question. Understanding the question to be answered will dictate the text data that needs to be acquired. For example, let’s say the marketing analyst is interested in knowing how customers feel about the quality of their products. Once the question is defined, the analyst can collect text data using multiple sources, such as online reviews downloaded from the company’s e-commerce site or scraped from a third-party website, and from call center transcripts, surveys, or social media posts. Typically, posts from social media, such as Twitter and Facebook, are downloaded using application programming interfaces (APIs). An API is a programming code that enables the user to connect one application program to another to access and download data seamlessly. Once the text data is collected, it needs to be combined and uploaded into the text analytics software program. Most software such as Python and R enable the user to define all text data into a corpus (“body” in Latin). A corpus refers to a collection of a large body of text that is ready to be preprocessed. Step 2: Text Preprocessing The second step includes processing and cleaning the data. The data often contains punctuation, variations in capital and lowercase letters, or special characters such as % or @. As a result, the data must be preprocessed. This step reduces noise in the data, creating consistency and removing special characters. Preprocessing prepares data for the text exploration and modeling. Critical concepts under text preprocessing include the following: Tokenization Tokenization is the process of taking the entire text data corpus and separating it into smaller, more manageable sections. These smaller sections are also known as tokens. A token is known as the basic unit of analysis in the text analytics process. Text analytics computer programs have their own predefined delimiters such as spaces, tab, and semicolons to divide the text into tokens. For example, the sentence “I love bread” would need to be separated into three tokens: “I,” “love