CASE STUDY DETAILED INSTRUCTIONS

‘What's the Next Big Thing?’: Netflix and Analyzing Data for Insights
N.B.: Before starting this assignment, read the case study, ‘What's the Next Big Thing?’: Netflix and
Analyzing Data for Insights.

Your assignment
Put yourself in Zach Joel’s shoes. You’re hoping to glean high-level insights into the key questions
you’ve agreed to explore for Netflix’s head of content marketing. Consider online self-service tools from
Netflix’s digital platform partners – such as Google and Facebook – and open-source datasets from
IMDb.

Your objective is to find interesting patterns that offer insights into your key questions. The output from
this exercise should be a collection of exploratory, data-driven charts and tables that identify areas for
further exploration and refinement as your data story takes shape.

Follow these steps to complete your assignment:

Collect & clean IMDb data

1. Download the seven files found at IMDb’s facilitated download site here
2. Unzip files by clicking on them
3. For simplicity’s sake, leave the files in your “Downloads” folder

Analyze IMDb data

4. Launch R and use the script found here to conduct a thorough analysis of the data
5. After completing the analysis, conduct further analysis of csv files you’ve exported to your
desktop (in the last step of the R script) using one or more tools:
a. Spreadsheet tools like Excel or Google Sheets
b. Presentation tools like Powerpoint or Google Slides
c. Dataviz tools like Tableau
6. Produce a series of data-based, exploratory data visualizations that reveal some interesting
patterns in the IMDb data
Use Google Trends to identify key areas of consumer interest (and geographic relevance)
7. Open a web browser and navigate to trends.google.com
8. Enter a search term or topic related to your insights in the query box (labeled, “Enter a search
term or topic”) and hit enter
a. Remember these important rules of search on Google Trends with example queries
related to the topic of tennis shoes
i. Tennis shoes: Searches will include tennis AND shoes, in any order, and possibly
with other search terms
ii. “Tennis shoes:”:Searches that include tennis shoes – that is, the exact search
terms inside the quotation marks
iii. Tennis - shoes: Google Trends includes only searches for tennis – not for shoes
iv. Tennis + shoes: Google Trends includes tennis OR shoes, in any order, and
possibly with other search terms
v. Sneakers + sneekers + sneakerss: Google Trend searches include alternative
spellings and aren’t case-sensitive
b. Note that Google Trends provides access to a largely unfiltered sample of Google search
requests that's anonymized (no one is personally identified), categorized (determining
the topic for a search query), and aggregated (grouped together)
9. Interpret the results of your query
a. Google Trends will return a “search index” for the term(s) you’re exploring – not a
number that’s the actual search volume for term(s)
i. For example, “100” on the Google Trends report doesn’t mean only 100
searches were conducted on that day for the term(s) you’re exploring
ii. There could have been hundreds, millions, or even billions of searches for that
term(s) on that day, but Google keeps those figures confidential
b. Google Trends normalizes search data to simplify comparisons between terms
c. Results are normalized to the time and location of a query using this process:
i. Each data point is divided by the total searches of the geography and time range
it represents to compare relative popularity (otherwise, places with the most
search volume would always rank highest)
ii. The resulting numbers are then scaled on a range of 0 to 100 based on a topic’s
proportion to all searches on all topics
iii. Different regions that show the same search interest for a term don’t always
have the same total search volumes
 d. Google Trends data reflects searches people make on Google every day, but it can also
reflect irregular search activity, such as automated searches or queries that may be
associated with attempts to spam search results
i. While Google has mechanisms to detect and filter irregular activity, Google
Trends may retain these searches as a security measure. Filtering them from
Google Trends would help those issuing such queries to understand how Google
has identified them, making it harder to keep such activity filtered out from other
Google Search products where high-fidelity search data is critical
ii. Given this, those relying on Google Trends data should understand it's not a
perfect mirror of search activity
e. Google Trends filters out searches, including:
i. Those that few people make: Trends only shows data for popular terms, so
search terms with low volume appear as “0”
ii. Duplicate searches: Trends eliminates repeated searches from the same person
over a short period
iii. Special characters: Trends filters out queries with apostrophes and other special
characters
f. Note that if you enter a search term using non-Latin characters, you only see data from
countries or regions that use those characters
i. For example, if you enter “ねこ,” the Japanese characters for “cat,” you don’t see
much data for the U.S. since many people in the U.S. use “cat” as their search
term
ii. To compare searches for Japanese characters for “cat” to searches for “cat”in
English, search for both terms by combining them with a “+” sign, like “ねこ +
cat”
10. Refine your search to add context and deepen your insights
a. Add a new term (or terms) to compare to your original term
i. Inside the Topics box, click + Add term and add another search term or terms
ii. To remove a term, hover over the search term box and click X
b. Change the geographic location of your results in one of two ways:
i. Change the geographic scope of all your results by selecting a new location
from the drop-down list below the search box (the location will default to your
present geography)
ii. Change the geographic scope of a particular term by clicking on the three dots
 on right side of the search box,select, “Change filters,” and then choose a
different location for that term
c. Change the period for your results in one of two ways:
i. Change the period of your results by selecting a new time frame from the
drop-down list below the search box (it will default to the past 12 months)
ii. Change the geographic scope of a term by clicking on the three dots to the right
side of the search box, select, “Change filters,” and then choose a different
period for that term
d. Depending on your needs, define your search words as terms or topics by clicking on
the term and making a selection
i. Terms: Search terms show matches for all terms in your query, in the language
given
1. If you search the term,“banana,” results include terms like “banana” or
“banana sandwich”
2. If you specify “banana sandwich,” results include searches for “banana
sandwich,” as well as “banana for lunch” and “peanut butter sandwich”
ii. Topics: Topics are a group of terms that share the same concept in any
language. Topics display below search terms.
1. If you search the topic, “London,” your search includes results for topics
such as “Capital of the UK” and “Londres,” which is “London” in
Spanish
e. If you’re using Trends to search for a word that has multiple meanings, you can filter your
results to a certain category to get data for the right version of the word
i. For example, if you search for “jaguar,” you can add a category to indicate if you
mean the animal or the car manufacturer
ii. Under the search box, click “All categories,” then choose a category
f. You can filter your results to a certain search platform to get data for that platform
i. Under the search box, click “Web Search,” then choose a platform
1. Image Search
2. News Search
3. Google Shopping Search
4. YouTube Search
11. Explore results by region to gain geographic insights
a. When you search for a term in Google Trends, you’ll see a map that shows areas where
 your term is popular
i. Darker shades indicate where your term has a higher probability of being
searched
b. If you compare search terms, you’ll see a map of the world shaded according to the
term’s popularity
i. The color intensity represents the percentage of searches for the leading search
term in a particular region
ii. Search term popularity is relative to the total number of Google searches
performed at a specific time, in a specific location
c. Hover over a region to get details on search volume in that region
i. To the right of the map, you’ll also see a list of regions or cities ranked according
to the term’s popularity
d. To view a metro, which are geographical areas that generally correspond to metropolitan
areas, follow the steps below (N.B.: Currently, Google Trends only provides metros for
some countries):
i. In the Regional interest section, click the U.S. on the map
ii. Click a state on the map
iii. Click a metro area on the map
iv. If a region on the map isn’t highlighted, it doesn’t mean there’s no interest.
Google Trends data is adjusted, so the term may be used in that region, but it’s
more popular in other regions.
12. Find related searches to expand your results
a. When you search for a term in Trends, you’ll see searches related to your term in the
related searches sections at the bottom of the page
b. If you're comparing multiple search terms, locations, or time ranges, you can see related
top searches by selecting the tab for your term
i. Top searches: Top searches are terms that are most frequently searched with the
term you entered in the same search session, within the chosen category,
country, or region. If you didn’t enter a search term, top searches overall are
shown.
ii. Rising searches: Rising searches are terms that were searched for with the
keyword you entered (or overall searches, if no keyword was entered), which had
the most significant growth in volume in the requested period. For each rising
search term, you’ll see a percentage of the term’s growth compared to the
 previous period. If you see “Breakout” instead of a percentage, it means that the
search term grew by more than 5,000%.
13. You can export Trends data to see a comprehensive list of search data as CSV files:
a. In the top right of the chart, click Download
b. Open the file using a spreadsheet application, like Google Sheets

Use Facebook Audience Insights to size audiences and collect demographic details
14. Navigate to www.facebook.com/business/insights/tools/audience-insights
15. Click on the “Go to Audience Insights” button
a. You’ll need a Facebook account to use Audience Insights. Facebook will ask you to log
in to your account if you’re not already logged in.
b. You may also need a Facebook Business Page, which you can easily create (for a real or
pretend business) by following these instructions
c. Facebook may ask you to “Choose an Audience to Start”
i. If asked, choose “Everyone on Facebook”
16. Navigate to “Insights” on the left side panel, then “Audience” in the second panel
a. Make sure that you select “Potential audience” at the top of the page to see the entire
Facebook audience
17. Use the “Filter” tool to create an audience relevant to your IMDb analysis using the filters on the
left side of the Audience Insights report
a. Make sure to deselect your default location (usually your home country) so that you do
not limit your analysis
i. Netflix is, after all, a global operation and understanding where trends exist (and
where they don’t) is very valuable to the company
b. Important topics like movie genres and actors can be easily found by simply typing them
into the “Interest” search box
c. Use the additional selections to bring more texture to the interest trends you’ve identified
d. Note that Facebook is constantly updating the number of attributes they allow you to
use in the Filter tool and the list below represents the options that are available at the
writing of this guide (N.B.: Try to limit your selections inputs that clarify the interest so
that the audience demographics arise organically from the Facebook data)
i. Location (country, region, or city)
ii. Age (or age range)
iii. Gender
 iv. Interests
v. Languages spoken
18. Note that Facebook will create a demographic report of the audience you create
a. Facebook estimates how many people fit your audience criteria. This is an estimate of
the size of the audience that’s eligible to see an ad on Facebook. It’s based on your
targeting criteria, ad placements, and how many people were shown ads on Facebook
apps and services in the past 30 days.
b. This isn’t an estimate of how many people will actually see your ad, and the number may
change over time. It’s not designed to match population or census estimates.
19. Take note of the Top Pages and Locations (Top Cities, Top Countries)
20. Use this information to bring depth to your understanding of the audiences Netflix should
consider and related interests that Netflix should consider

Use Twitonomy & LIWC to sample broad consumer sentiment

21. Open a web browser and navigate to www.twitonomy.com
22. Sign in with your Twitter account (if the app asks you to do so)
23. Click on “Add a user”, “Add a list”, or “Add a keyword search” to begin analysis
24. Enter a account, list name, hashtag, or keyword you want to explore in the search box (based on
which category of search you selected in the previous line)
a. Think of titles, properties, actors, directors, and topic areas that are related to what you
know about Netflix’s opportunities and consumers
25. Refine your search to deepen your insights
a. Twitter supports a number of advanced search operators and filters that allow you to
customize your search query and find the exact tweets you’re looking for.
b. Here are a few of the best Twitter search operators and how to use them (with
examples):
i. Find one keyword OR another: First, Twitter doesn’t require an AND or +
operator to search for multiple keywords. So don’t include them. Just type
multiple keywords in your query and Twitter will return tweets that include those
terms. For example: social media metrics.
1. Sometimes, however, you might want to find tweets that include one
keyword or another keyword. Use the OR operator to separate those
terms and your report will include tweets that mention one or the other.
2. You can also chain together multiple keywords to create a more
 complex query. The OR operator will attach to the word that immediately
precedes it, very much like order of operations in algebra. For example,
social media metrics OR analytics will find tweets that mention social
media metrics or social media analytics, because the OR links to the
metrics and analytics terms.
ii. @Username queries: There are several ways to learn more about the reach of
tweets from a particular Twitter account, depending on the type of information
you’re looking for.
1. Tweets to, from and about an account (e.g., “kevinhartman”): Run a
report for a username but don’t include the @ symbol. This will return all
mentions and tweets (including retweets and replies), as well as all
tweets from that Twitter account. This is the most comprehensive set of
reach stats for a specific Twitter account.
2. Tweets to and about an account (e.g., “@kevinhartman”): Run a report
for a username and include the @ symbol. This will return all mentions of
an account, but not any tweets from that account. This report will let you
know how many people are talking about a certain Twitter account, and
the ways they’re talking about it (including all retweets, replies, and
mentions).
26. Use a tool like LIWC to conduct sentiment analysis on the Tweets that are returned
a. Download the results of your Twitonomy report into a .csv file by clicking on the
“Download & Print” icon at the top of the report data
b. Open the downloaded file in a spreadsheet like Google Sheets or Microsoft Excel
c. Find and copy the “Text” content under the header, “Tweets”
i. Note that the “Text” content is what was actually tweeted and that the other
data under the “Tweets” header -- Id, Screen_Name, Time, Retweets,
Impressions, URL –is metadata created by Twitter
d. Navigate to the LIWC homepage and paste the content into the box at the bottom of the
page next to the text, “GIVE IT A TRY”
e. In the drop-down menu labeled, “How would you classify this text (choose one):,” select
“Social media: Twitter, Facebook, blog” and then hit the “ANALYZE” button
i. Note that the free LIWC report caps its analysis at 500 words. If you entered
more than 500 words, only the first 500 words were analyzed. The paid-for
LIWC2015 report actually produces about 90 different output dimensions.
f. LIWC will return a report containing the following variables:
i. Word count (WC): The raw number of words within a file.
ii. Words per sentence (WPS): The mean number of words within each sentence
within the file
iii. Percentage of total words: Most of the LIWC output variables are percentages of
total words within a text. For example, imagine you have analyzed a blog and
discover that the Positive Emotions (or posemo) number was 4.20. That means
that 4.20 percent of all the words in the blog were positive emotion words.
iv. Analytical thinking (Analytic): The analytical thinking variable is a
factor-analytically derived dimension based on eight function word dimensions.
Originally published as the categorical-dynamic index, or CDI, the dimension
captures the degree to which people use words that suggest formal, logical, and
hierarchical thinking patterns. People low in analytical thinking tend to write and
think using language that is more narrative ways, focusing on the here-and-now,
and personal experiences. Those high in analytical thinking perform better in
college and have higher college board scores. To learn more about analytical
thinking, see Pennebaker, Chung, Frazee, Lavergne, and Beaver (2014).
v. Clout: Clout refers to the relative social status, confidence, or leadership that
people display through their writing or talking. The algorithm was developed
based on the results from a series of studies where people were interacting with
one another (Kacewicz, Pennebaker, Davis, Jeon, & Graesser, 2013). Note that
Clout is different from the LIWC2015 Power variable. Power or, more accurately,
need for power, reflects people’s attention to or awareness of relative status in a
social setting. You can have a confident leader who has no interest in other
people’s standing in the social hierarchy.
vi. Authenticity: When people reveal themselves in an authentic or honest way, they
are more personal, humble, and vulnerable. The algorithm for Authenticity was
derived from a series of studies where people were induced to be honest or
deceptive (Newman, Pennebaker, Berry, & Richards, 2003) as well as a summary
of deception studies published in the years afterwards (Pennebaker, 2011).
vii. Emotional tone (Tone): Although LIWC2015 includes both positive emotion and
negative emotion dimensions, the Tone variable puts the two dimensions into a
single summary variable Cohn, Mehl, & Pennebaker, 2004). The algorithm is built
so that the higher the number, the more positive the tone. Numbers below 50
 suggest a more negative emotional tone.
g. Note that the LIWC report will include a comparison of “YOUR DATA” to the “AVERAGE
FOR SOCIAL MEDIA: TWITTER, FACEBOOK, BLOG” to provide context to your sample

Use YouTube comments & LIWC to sample targeted consumer sentiment

27. Get a YouTube API key by following these instructions
28. Make sure the API key is tied to the @gmail account you use to sign-in to YouTube
29. Use this R script to pull key information about particular YouTube channels, including the videoID
for each of the channel’s videos
a. You can find the Channel ID for a YouTube Channel you want to analyze using the tool at
provided by Channel Picker
b. If you are not interested in collecting YouTube Channel-level data for insight into a
specific YouTube Channel (e.g., Paramount Pictures channel) and only wish to pull
YouTube comments for specific videos (e.g., a trailer for a specific Paramount Pictures
movie), then you can skip this step
30. Collect YouTube comments for the videos you’ve identified using the Apps Script function of
Google Sheets
a. The R script will provide the videoID for every video published by the channel you’ve
researched, but alternatively (or if you skipped the R script step) you can find the videoID
with help of this gude
b. You will find a video demonstration of a solid process here
i. To use this approach, you’ll need to copy and paste the script found at the
GitHub sites below into the Google Sheet Apps Script function
1. Code for Comments without Replies
2. Code for Comments with Replies
c. Alternatively, you can use this approach developed by Google
31. Once you have collected comments, copy and paste them into the LIWC tool to conduct
sentiment analysis in the same way you did with the Tweets you collected earlier

Complete your assignment

32. Pull together your collection of exploratory, data-driven charts and tables
a. Make note of the areas you’ve identified for further exploration and refinement as your
data story takes shape
b. Think about additional data (and data sources) you’d like to add to bring further clarity to
 your analysis or expand into interesting areas
33. Complete a write-up that answers Netflix’s question: “What’s the next Big Thing?” using the
analysis plan given to you in the Lab
a. Offer a clear recommendation that is clearly supported by findings in your analysis
b. Make sure to pull data into your analysis using multiple tools (e.g., those detailed in the
steps above) to acquire a deep and robust data asset that clearly links to the key
questions driving your analysis
c. Thoroughly analyze tidy, error-free data that result in creative insights supporting your
key questions