SOLUTIONS MANUAL
Essentials of Marketing Analytics
Chapter No. 01: Introduction to Marketing Analytics
Sample Answers to Discussion and Review Questions
1. What is marketing analytics?
Marketing analytics uses data, statistics, mathematics, and technology to solve marketing
business problems. It uses modeling and software to drive marketing decision making.
Today’s availability of large amounts of data, improvements in analytics techniques,
substantial increases in computer processing power, and affordability have made marketing
analytics more practical and available to a much larger audience. Until recently, many
organizations made decisions based on intuition or opinion versus data-driven knowledge.
Data analytics techniques provide an excellent opportunity to bridge the gap between
information and insights.
2. How are companies using marketing analytics to make strategic marketing decisions?
Companies increasingly need to differentiate products and services, optimize processes,
and understand the drivers for business performance, and marketing analytics can help
them do so. Marketers can use insights from analytics to increase company performance
through pricing, product development, channel management, marketing communications,
and selling. Restaurants have begun applying marketing analytics to optimize location
selection. The restaurant chain Roy Rogers Franchise Co. uses advanced analytics to
expand into new markets, determine their next site locations, and forecast sales. Their
machine learning platform integrates internal and external data to ensure restaurant
locations match the needs and wants of the geographical area.
3. Name several external data sources that might be helpful to marketers.
Google Dataset Search includes data from NASA, NOAA, Harvard’s Dataverse, GitHub,
Kaggle, and other sources. Social media content can be mined for words or phrases
associated with a product or company. Companies can gather data collected by
distribution channel partners. Commercial brokers collect and sell both public and private
data including socioeconomic status, health interests, and political views. The government
provides more than 200,000 datasets searchable by topic. The U.S. Census Bureau
provides data on population, economy, housing, and geography. Another dataset compiles
consumer complaints about financial products and services. There are demographic
statistics by ZIP code, gender, ethnicity, and citizenship. A dataset collects fruit and
vegetable prices for many commonly consumed items through the Department of
Agriculture. There is also a dataset for tax return data by state and ZIP code level.
,4. How might a company use structured and unstructured data to better understand
customers?
Structured data includes numbers, dates, and text strings stored in a clearly defined
structure of rows and columns. Unstructured data includes text, images, videos, and sensor
data and does not fit well into a table format. Unstructured data required advanced
analytics techniques such as AI to prepare and analyze. Companies can combine
unstructured and structured data to better understand their customers. Combining
information from conversations on support sites with information from individual customer
accounts deepens customer understanding. Technology has improved the combination of
these kinds of data structures but user knowledge are lagging. AI and machine learning
will be increasingly used to help combine and analyze this data.
5. Define a target variable.
Variables are characteristics or features that pertain to a person, place, or object.
Marketing analysts explore relationships between variables to improve decision making.
Say an analyst is investigating the relationship between two variables: store lighting and
total sales per customer. The store lighting is considered the independent variable or what
influences or drives the dependent, target, or outcome variable – sales per customer. The
target variable is sales per customer and variables other than lighting may be affecting
sales per customer. Companies can use multiple variables at the same time as inputs to
systems that process data and use it to predict target variables.
6. Discuss the difference between supervised and unsupervised learning.
Both are types of algorithms used to address business problems. Supervised learning
suggests the target variable of interest is known and available in a historical dataset. The
historical dataset, or labeled data, is divided into a training dataset, a validation dataset, and
an optional testing dataset. Unsupervised learning has no previously defined target
variable. The goal of unsupervised learning is to model the underlying structure and
distribution in the data to discover and confirm patterns.
7. What are the steps of the marketing analytics process?
The 7-step marketing analytics process is iterative and continuously evolves to develop and
manage improvements in the modeling cycle. Each step plays an important role in
achieving a successful outcome. The steps in the process are:
• Step 1: business problem understanding
o A key element here is to question whether the problem the business is
presenting is, in fact, the correct problem
• Step 2: data understanding and collection
, o This step includes examining internal and external databases and talking
with key data owners and stakeholders
o It is also determined if the identified problem is the actual problem or a
symptom of an underlying problem.
• Step 3: data preparation and feature selection
o Data in different formats is cleaned and combined in this step
o Understanding the meaning of each variable and its unit of analysis is an
essential task in this step
• Step 4: modeling development
o In this step, the analyst selects the method to use
o Different models should be tried to identify the one providing the best
accuracy, speed, and quality
• Step 5: model evaluation and interpretation
o This step evaluates the model to identify the algorithm providing the best
solution
• Step 6: model and results communication
o The analyst needs to present the model in a way that other people can
understand, particularly management
• Step 7: model deployment
o The model must now be implemented and run on real-time records to offer
decisions or actions
Critical Thinking and Marketing Applications
1. Visit www.data.gov. Click on Consumer, then click on Data. How many datasets are
currently located on this website for free? Select one dataset and develop a scenario
where the data might be helpful for a marketing manager. Discuss how exploring the
data could guide the marketing manager in making more informed decisions.
There are 218,078 datasets on the site and 113 with the topic of “Consumer,” as of
November 2020. Suppose you work for a company that supplies tents for weddings and
other gatherings. Top management is thinking about branching out into making awnings
for patios and decks. You locate the Characteristics of New Housing dataset which
supplies you with the information that a third of newly constructed single-family homes
included a patio and a porch while only eight percent had no outdoor features. You could
explore the data further looking for income levels in your area or home improvements.
You could also look at weather patterns. Sometimes exploring the data may spark thoughts
on other products you would not have thought of otherwise. For example, in looking at the
new construction for single-family homes, you notice nearly 90 percent have a central air
conditioning unit. Maybe adapt the tents you now produce to fit various air conditioning
, units for winter storage. Or pool covers. Looking at the data can confirm theories or bring
to light customer aspects you had not considered.
2. Develop two questions that an airline company might be interested in answering.
Describe types of unstructured and structured data that might be important to
answering the questions. What data sources might be helpful?
How many times a month does a customer report lost baggage and does this affect the
customer from flying with our airline in the future?
The airline can use structured data to find how often in a month a customer reports lost
luggage, this should be an actual number. Repeat customers are tracked using frequent
flyer programs or other customer loyalty programs, which would be structured, accessible
data. As for unstructured data, the airline could use video of the customer interaction when
reporting the lost luggage to gauge the customer’s emotional reaction, listening for threats
of “I’ll never fly this airline again.” The company could also mine social media sites for
search terms including their name and lost luggage for any rants (or compliments) about
their airline. The airline could also collect primary data by surveying customers in the
frequent flyer program to determine how many have reported lost luggage with the airline.
Meshing the internal data and external data may require advanced analytics techniques
using AI to prepare and analyze.
How many personal injuries occurred during the boarding and departing processes each
month and how can we increase safety during these times?
Structured data would include all recorded injuries during boarding and departing.
Unstructured data could be interviews conducted with the injured customer, video footage
of the accident, and written injury reports. Analysts charged with increasing safety during
these times can use prior incidents to predict, and prevent future injuries. An external
source that may prove useful here is the NTSB Aviation Safety News, available as an RSS
feed.
