You are on page 1of 30

Social Media

Geoforensics:
An Exploratory Analysis
Sy Banerjee,
Associate Professor
School of Management
University of Michigan, Flint,
syban@umflint.edu
Fareena Sultan,
Professor
DAmore McKim School of Business
Northeastern University, Boston, MA, USA
f.sultan@neu.edu
Presented at EMAC 2016, Oslo, Norway
May 25, 2016

CONTENTS
BACKGROUND
KEY RESEARCH QUESTION
DEFINITION OF SOCIAL MEDIA GEOFORENSICS
DATA COLLECTION AND SAMPLE
TWEET CONTENT CLASSIFICATION AND CODING
ANALYSIS
IMPLICATIONS
CONTRIBUTIONS

SOCIAL MEDIA SHARING


Social Media sharing can have promotional
value in Integrated Marketing Communications
(Mangold and Faulds 2009)
Can capture attention of consumers, trigger
likes, comments and brand popularity (Vries,
Gensler and Leeflang 2012)
Can act as Product Evaluation and
Recommendation to friends (Chen, Fay, Wang
2011)

CONTENT ANALYSIS OF
TWEETS
Tweet sentiments can be reflective of electoral
public opinion (O'Connor, Balasubramanyan,
Routledge and Smith, 2010)
Can help predict brand sentiments, sales
performance (Zhang, 2014) and movie success
(Jain 2013)

LOCATION-BASED SOCIAL
NETWORKS
Apps like Foursquare, Gowalla help individuals share
personal experiences with friends real time
Increasing user bases (Foursquare: 60 million, 2015 August)
indicates increasing number of people that are comfortable
sharing personal location
Based on location data, several companies now offer
location based social media monitoring such as Geofeedia,
Welink, Sysomos, Momentfeed, Venuelabs and Local
Measure
5

LOCATION-BASED SOCIAL MEDIA


MONITORING
LISTENING AND ANALYSIS

In-store
retail
aisles
Neighborhoods

Key Research
Question
Given the exponential rise in social media
sharing from a variety of physical
locations, how does context affect usage?
What can marketers learn from these
context-specific behaviors?

SOCIAL MEDIA
GEOFORENSICS:
WHAT IS IT?
Analyzing physical context-specific information
from Location Based Social Networks to design
new metrics that help marketers make sense of
mobile consumers.

What does Social Media


Geoforensics entail?

Identifying social media platforms revealing location

Connecting context variables that help explain shared


information

Coding shared content

Understanding user characteristics based on sharing


patterns

Analyzing user and context effects on sharing behavior


9

Drawing implications for businesses and marketers from


results

Overall Data
Collection and
Analysis Process

10

DATA
Total 2412 tweets, From 1740 locations, By 323
users
193 men 1435 tweets (7.4 per user)
130 women 977 tweets (7.5 per user)
City Locations (1 mile radius)- Chosen for high
traffic
Chicago (Union Station)
New York (Central Park)
Cambridge (Boston, Harvard Square)
DC (Union Station)
Seattle (Capitol Hill)
San Francisco (Market Street Twitter Office)

11

12

CONTENT ANALYSIS
TWEET CATEGORIZATION

13

TWEET CATEGORIZATION
DESCRIPTIVES

14

RESEARCH QUESTIONS
Are all venues created equal? Are users more likely to
check into certain locations more than some others?
Once checked in, how does venue and user type
affect the content of tweet (what they tweet about)?
What kind of venues are checked into by different
genders on different days of the week?
Which venues do different genders go to with
more/less friends?
15

NEW METRICS
USER CHARACTERISTIC:
VENUE VARIABILITY: HOW FLEXIBLE ARE THE USERS IN CHECKING IN AT
MULTIPLE TYPES OF VENUES (FOOD, ARTS & ENT, RETAIL, TRANSPORT)
VENUE CHARACTERISTICS
VENUE IMPORTANCE (SHARE OF CHECK INS WITHIN CATEGORY)
= NUMBER OF CHECKINS IN VENUE /TOTAL NUMBER OF CHECKINS IN CATEGORY
VENUE TIMING = DAY OR DAYPART EFFECTS
VENUE SOCIAL IMPORTANCE = (AVERAGE NO OF FRIENDS ACCOMPANIED)
INWARD/OUTWARD DIRECTEDNESS = (NUMBER OF PERSONAL TWEETS /
NUMBER OF VENUE-RELATED TWEETS)
16

COMPUTING VENUE
VARIABILITY

Keeping number of check-ins constant as more categories are


checked into, the Unalikeability Index, or Venue Variability (VV)
increases. So we computed the VV of users and created three
groups for each gender
Men (%) Women (%)
Situated (least flexible)
16 11
Moderate
50
53
Ubiquitous (most flexible)
34 36

17

SHARE OF CHECK INS BY VENUE AND


VENUE VARIABILITY FOR MEN

18

SHARE OF CHECK INS BY VENUE AND


VENUE VARIABILITY FOR WOMEN

19

VENUE IMPORTANCE

3.3%

0.5%

0.8%

0.5%
20

0.8%

VENUES AND TIMING BY


GENDER

21

NO. OF FRIENDS BY VENUE AND


GENDER

22

TWEET CONTENT BY VENUE,


VENUE VARIABILITY AND GENDER
Inward-Outward Directedness
= (Number of Personal Tweets / Number of Venue-Related
Tweets)

23

MANAGERIAL ACTIONS
-1
Segment/ target for cross promotions, new experiences
Michael checks in at

Clarke checks in at

Who will you target?

24

MANAGERIAL ACTIONS -2

What is the check-in share of your brand within your product categor

If you wanted to engage in Geo Conquesting, which


competitors customers will you target?

25

MANAGERIAL ACTIONS -3
What are the prime times and categories for crosspromoting your brand or category?
On Tuesdays/
Thursdays, if
targeting men, use
food coupons, for
women use both
food and retail

26

MANAGERIAL ACTIONS
-4 for your category?
How do you target groups
If in Arts & Entertainment or
Outdoors, men are with more
friends.
If in Retail or Food, women are in
company of more friends

27

MANAGERIAL ACTIONS -5
What kind of information can you mine from your
venue?

When the I/O ratio is > 1, there is more inner directed


content and you can learn more about customers personal
lives and needs
When the I/O is < 1, there is more outer directed content,
and you can find out more about customers experiences,
and what aspects of the products or services they pay

28

Contributions

Conceptualizing and computing venue


variability
New way to categorize tweet content
Metric connecting context to content =
Inward Outward directedness

29

QUESTIONS?

You might also like