Using Customer Data for Analytics

CUSTOMER RELATIONSHIP MANAGEMENT

Professor Constant Cheng, PhD

2024
Learning Objectives

1. Understand the role of analytics for strategy and tactics

2. Describe how analytics supports customer management
throughout the customer journey in terms of marketing, sales,
and service.
3. Describe the role of data mining for: describing and visualizing,
classification, estimation, prediction, affinity grouping and
clustering.
4. Describe the types of analytics that apply to structured,
unstructured, and Big Data.
5. Describe the role of Artificial Intelligence (AI) in analytical CRM.
What are customer data analytics for?

Customer analytics help companies predict future trends, uncover associations, and
categorize customers for better strategic decision-making.

Predicting Future Trends:

•Analyzing past customer behaviors to forecast future actions,
improving operational efficiency and effectiveness.

•Uncovering Associations:
Identifying surprising connections between variables that can lead to
valuable insights for achieving CRM goals.

•Customer Segmentation:
Grouping customers with similar characteristics to make targeted
decisions on whom to serve and how to approach them.
Analytics for CRM Strategy and Tactics

Analytics for CRM strategy and tactics involve using data to achieve long-term strategic
goals and short-term tactical actions, focusing on revenue growth, cost reduction, and
customer loyalty enhancement.
 Analytics Throughout the Customer Journey

Analytics throughout the customer journey involve using different types of analysis to
support strategies for customer acquisition, retention, and development, each tailored
to the specific phase of the journey.

Customer acquisition strategies involve identifying and qualifying potential customers,

typically using lead scoring based on various market, organizational, personal,
relational, and behavioral attributes, with higher scores indicating better prospects.
 Analytics for Structured Data

Structured data is data that fits neatly into data tables and includes discrete
data types such as numbers, short text, and dates. Structured data are either
discrete but qualitative/categorical (divided into groups), or discrete
with whole number or continuous(has decimal) and quantitative data
that can take any value over a particular time interval).
•Nominal Data (categorical data that are unordered):
Categorical data used for classification or identification without any order, such as
customer IDs or gender.
•Ordinal Data (categorical data that are ordered:
Ranked data that indicate order but not the magnitude of difference, like a list of
customers ranked by sales, or degree of customer satisfaction.
•Interval Data (continuous data):
Data that show both order and the exact distance between values but have no true
zero point, such as customer perception of product quality. Zero in temperature is just
a particular point.
•Ratio Data (interval data with true zero):
Data with all properties of the other types, plus a true zero point, allowing for
comparison of ratios, like customer income or expenditure.
 Analytics for Unstructured Data

Analytics for unstructured data involve advanced techniques to extract meaningful insights from
non-structured sources like text, audio, and video, enabling businesses to better understand
customer behavior and trends.

Automatic Routing:
•Uses analytics to automatically route emails, detect spam, and manage other communication
flows efficiently.
Root Cause Analysis (RCA):
Analyzes customer service or complaint records to identify the underlying causes of issues.
Sentiment Analysis:
Evaluates customer feelings and opinions about products, services, or brands to gauge overall
sentiment (positive, negative, or neutral).
Text Analytics:
•Converts unstructured text into structured data to reveal hidden themes and concepts,
improving predictive models and decision-making.
Trend Analysis:
Tracks changes in specific entities over time to understand patterns in customer behavior or
product performance.
Sentiment Analysis
 Big Data Analytics

Big Data analytics involves analyzing large, fast, and diverse data sets to extract valuable insights
that can improve decision-making and customer experiences.

Volume:
Refers to the vast amount of data generated, which exceeds the capabilities of traditional databases.

Variety:
Involves the diverse types of data, including structured and unstructured formats like social media
posts, videos, and images.

Velocity:
Describes the high speed at which data is generated and needs to be processed, often in real-time.

Veracity:
Concerns the quality and accuracy of the data, ensuring that insights are reliable and actionable.
 Example of Data Analytics
Walmart
https://www.projectpro.io/article/how-big-data-analysis-
helped-increase-walmarts-sales-turnover/109#toc-1

Structured Data:
Example:
Walmart uses structured data in its POS systems to track sales, manage inventory,
and analyze purchasing patterns through product codes, prices, and timestamps.

Unstructured Data:
Example:
Walmart analyzes unstructured data from social media and customer reviews to
understand customer sentiment and improve marketing and service strategies.

Big Data Analytics:

Example:
Walmart uses big data analytics to predict demand and optimize inventory by
analyzing large datasets, including purchase history and weather patterns.
Tech Architecture & Ecosystem
Walmart

The main objective of leveraging big data at Walmart is to optimize the shopping experience
for customers when they are in a Walmart store or browsing the Walmart website
Three Ways to Generate Analytic Insights

CRM users can generate analytical insights through three main methods: standard
reports, OLAP (online analytical processing), and data mining.

•Standard Reports:
Predefined or customizable reports that provide users with essential customer-
related information, such as sales metrics or call center activity.

•OLAP (Online Analytical Processing):

Enables ad hoc data analysis through techniques like slice-and-dice, drill-down, and
roll-up, helping users explore data from different perspectives to uncover deeper
insights.

•Data Mining:
Utilizes advanced statistical techniques to discover patterns, correlations, and trends
in large datasets, providing predictive insights and decision support.
Standard Reports
OLAP (Online Analytical Processing)
OLAP tools to examine sales performance by drilling down into sales data for a specific product category, such as
Food and Beverages, then by gender, revenue, quantity, etc.
Data Mining

Walmart uses big data analytics to enhance customer experience and optimize
operations through social media insights, personalized recommendations, inventory
management, and improved checkout processes.

Social Media Analytics:

•Analyzing social media data to understand customer preferences
and inform product decisions.

Social Genome:
•Combining social and proprietary data to provide personalized
product recommendations.

Shopycat Gift Recommendation:

•A Facebook app suggesting gifts based on users' social data.

Predictive Inventory Management:

•Forecasting demand to optimize inventory and reduce overstock.

Improved Checkout Process:

•Using analytics to streamline checkout options and staffing.
Data Mining I

Data mining in CRM involves applying descriptive and predictive analytics to large data
sets to support marketing, sales, and service functions, using techniques like
classification, estimation, and clustering.

Classification:
•Categorizing data into predefined groups to predict outcomes, such as
identifying likely customers to respond to a promotion.
Estimation:
•Predicting numerical outcomes, like forecasting a customer’s lifetime value.
Prediction:
•Anticipating future behaviors, such as predicting which customers are likely
to churn.
Affinity Grouping:
•Identifying relationships between items, such as products often purchased
together.
Clustering:
•Grouping similar data points, such as segmenting customers based on
buying patterns.
Description and Visualization:
•Summarizing data to highlight key patterns and insights through visual
tools.
Data Mining II

Directed (Supervised) Data Mining:

•Using specific input data to predict a defined outcome, like
forecasting customer responses to a marketing campaign.

Undirected (Unsupervised) Data Mining:

•Exploring data without predefined outcomes to discover new
patterns, such as uncovering unexpected customer segments.
Directed Data Mining Techniques I

Decision Trees:
•Predict outcomes by splitting data into groups based on decision rules.
Example: A bank uses a decision tree to decide on loan approvals by
evaluating factors like credit score and income.

Logistic Regression (estimate the probability of an outcome):

•Assesses the impact of variables on a categorical outcome, a S curve between 0 and 1,
like purchase likelihood.
Example: An online retailer predicts which customers will respond to a
marketing email based on their past behavior.

Multiple Regression(a statistical procedure that uses a regression equation to predict

the value of a dependent variable based on multiple independent variables.)
•Explains a continuous outcome using multiple predictor variables.
Example: A streaming service analyzes how subscription cost and content
availability influence subscriber numbers.
Directed Data Mining Techniques II

Discriminant Analysis (DA) to analyze data when the dependent variable is categorical
and the independent variable is interval in nature): Classifies data into groups based on
distinguishing variables.
•Example: A retailer categorizes customers into value segments based on their spending
and purchase frequency.

Neural Networks:
•Machine learning models that predict outcomes by learning from data.
•Example: A telecom company predicts which customers might leave using patterns in
their service usage and complaints.
Undirected Data Mining Techniques I

Hierarchical Clustering:
Groups records into clusters by progressively merging them until one super-cluster remains.

Example: A dendrogram (tree diagram) is used to cluster markets based on sales data.
 Undirected Data Mining Techniques II

K-Means Clustering:
Clusters records into a predetermined number of groups (k) by minimizing variance
within clusters and maximizing distance between them.

Example: Clustering customers into three distinct segments based on purchasing behavior.
 Undirected Data Mining Techniques III

Two-Step Clustering:
Combines hierarchical and k-means clustering, suitable for large datasets with mixed data types.

Example:
A retail company wants to segment its customers to target marketing campaigns more effectively. They
use two-step clustering on data including demographics, purchase behavior, and engagement metrics.

Steps:
1.Initial Clustering (K-Means):
Group customers into broad clusters, such as high-income frequent shoppers, middle-income
occasional shoppers, and low-income infrequent shoppers.
2.Refinement (Hierarchical Clustering):
Further segment these groups based on additional factors, like engagement levels (e.g., website
visits, loyalty participation).
Example: High-income frequent shoppers might be split into highly engaged vs. less engaged
customers.
3.Implementation:
Use these refined segments to tailor marketing efforts, like offering premium product promotions
to highly engaged high-income customers and discounts to bargain hunters.
Undirected Data Mining Techniques IV

Factor Analysis:
Reduces data complexity by identifying underlying variables that explain observed patterns.

Example:
A company wants to improve its customer relationship management (CRM) strategy by
understanding what drives customer satisfaction and loyalty. They gather data on
various customer-related variables, such as:
• Response time to customer inquiries
• Quality of customer support
• Ease of use of the company's website
• Personalization of communication
• Variety of product offerings
• Pricing competitiveness

By applying factor analysis, the company discovers that these variables can be grouped
into three main factors:
1. Service Quality: Includes response time and quality of customer support.
2. User Experience: Comprises ease of use of the website and personalization of
communication.
3. Product and Price: Encompasses variety of product offerings and pricing
competitiveness.
 Artificial Intelligence (AI), Machine Learning (ML), & Deep Learning (DL)

AI, ML, and DL are advanced computing methods that evolved to make machines more
intelligent, with applications in CRM for predicting customer behavior using large
datasets and complex algorithms like Artificial Neural Networks (ANNs).

1.Artificial Intelligence (AI):

The broader concept of creating machines capable of performing tasks that typically require
human intelligence.
•Example: AI powers virtual assistants like Siri to understand and respond to user queries.
2.Machine Learning (ML):
A subset of AI that enables machines to learn from data and improve their performance over time
without explicit programming.
•Example: Netflix uses ML to recommend shows based on your viewing history.
3.Deep Learning (DL):
A subset of ML involving neural networks with many layers that can learn complex patterns from
large amounts of data.
•Example: DL models help Facebook automatically tag friends in photos by recognizing faces.
4.Artificial Neural Networks (ANNs):
A specific ML model inspired by the human brain that processes inputs through layers to make
predictions.
•Example: ANNs are used in CRM to predict customer churn by analyzing behavior patterns from
large datasets.
Example of AI, ML and DL

Amazon aims to enhance customer experience, predict customer behavior, and optimize
operations by leveraging advanced technologies across its e-commerce platform .

Artificial Intelligence (AI):

•Example: Amazon uses AI in its Alexa virtual assistant to provide personalized shopping
experiences. Customers can ask Alexa to order products, check delivery statuses, or
recommend items, making shopping more convenient and engaging.
2. Machine Learning (ML):
•Example: Amazon employs ML algorithms to analyze customers' browsing and purchase
history. This allows Amazon to offer personalized product recommendations on its website
and through email, increasing the likelihood of additional purchases.
3. Deep Learning (DL):
•Example: Deep learning models are used by Amazon to analyze and understand customer
reviews and feedback. By processing vast amounts of unstructured data, these models can
detect sentiment, identify emerging trends, and improve product recommendations or
customer support strategies.
4. Artificial Neural Networks (ANNs):
•Example: Amazon uses ANNs to predict customer churn by analyzing patterns in customer
behavior, such as changes in purchase frequency, product returns, and engagement with
Amazon Prime services. This helps Amazon proactively address potential churn by offering
targeted promotions or improved services to retain these customers.
 Class Exercise
Nike

Break the class into teams.

You have 45 minutes to find the answers.
Then, select a team spokesperson to present your findings .
Instructions:
1.Choose a Company:
• Select a large scale and well-known company that interacts directly with customers, such as
Netflix, Walmart, or Starbucks.
2.Analyze the Current CRM Strategy:
• Research how the chosen company currently manages its customer relationships. Consider
their customer service approach, marketing strategies, and personalization efforts.
3.Identify Opportunities for AI, ML, DL, and ANNs:
• Propose ways the company could implement AI, ML, DL, and ANNs to enhance their CRM
strategy.
• Think about how these technologies could be used to improve customer service,
personalize marketing, predict customer behavior, and optimize operations.
4.Provide Specific Examples:
• For each technology (AI, ML, DL, ANNs), provide a specific example of how it could be
applied in the company’s CRM strategy.
• Explain the expected outcomes and benefits of implementing these technologies.
5.Present Your Findings