Upload
dangliem
View
215
Download
2
Embed Size (px)
Citation preview
Social Media Intelligence
Wendy W. Moe [email protected]
www.wendymoe.com @wendymoe
David A. Schweidel [email protected]
www.dschweidel.com @dschweidel
• The challenge facing organizations
• Social Media Intelligence – Understand the science of opinion and the implications for
social media data and metrics – Measure what matters (not what’s easy to measure)
What is Social Media Intelligence?
2
Building Social Media Intelligence
3
WHAT?
WHERE?
WHO?
“Online Product Opinions: Incidence, Evaluation and Evolution,” Marketing Science, 2012, with David A. Schweidel
“The Value of Social Dynamics in Online Product Ratings Forums,” Journal of Marketing Research, 2011, with Michael Trusov
“Sentiment and Venue Choice in Social Media Posting Behavior,” under review at JMR, with David A. Schweidel
“Drivers of Social Media Rebroadcasting: Investigating the Role of Message Content and Influencers,” under review at JMR, with Yuchi Zhang and David A. Schweidel
INTE
GRAT
ION
Soci
al M
edia
Inte
llige
nce
by
Wen
dy W
. Moe
and
Dav
id A
. Sch
wei
del,
Cam
brid
ge U
nive
rsity
Pre
ss, c
omin
g Fa
ll 20
13
4
WHAT do we say?
Social Media Posting Decisions
Opinion / Brand Evaluation
Do I post? What do I post?
Where do I post?
SentimentProduct
Attribute
Venue FormatDomain
5
Rating Trends
Variance
Average
Activists
Post frequently Attracted by lack of consensus More negative Variance and volume make them more negative
Low Involvements
Post infrequently Deterred by lack of consensus More positive Variance and volume make them more positive
6
Implications
7
What should we do?
• If you’re listening, don’t over-react. The ratings environment does not necessarily reflect the opinions of the entire customer base
• “Listen to ALL the American people” – Encourage diversity of opinion – Monitor different venues / data sources
8
9
WHERE do we post?
Social Media Posting Decisions
Opinion / Brand Evaluation
Do I post? What do I post?
Where do I post?
SentimentProduct
Attribute
Venue FormatDomain
10
What affects posted sentiment?
General Brand Impression (GBI)
Venue Venue-specific dynamics Message topic
11
Does it Matter Where We Listen?
0%
20%
40%
60%
80%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Dist
ribut
ion
of C
omm
ents
Observation Month
Blog
Forum
Microblog
Other
0%
20%
40%
60%
80%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Prop
ortio
n of
Pos
itive
Com
men
ts
Observation Month
Blog
Forum
Microblog
Aggregate
12
From a sample of social media comments pertaining to an enterprise software brand (courtesy of Converseon)
Plus, sentiment metrics vary depending on topic
13
GBI vs. Offline Survey • Potential for GBI as a lead
indicator
• Correlation with survey (t) – GBI = .376 – Avg sentiment =.008 – Blogs = .197 – Forums = -.231 – Microblogs = .394
• Correlation with survey (t-1)
– GBI = .881 – Avg sentiment = .169 – Blogs = .529 – Forums = .213 – Microblogs = .722
8.75
8.8
8.85
8.9
8.95
9
9.05
9.1
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
1 2 3 4 5 6 7 8 9 10
Aver
age
Surv
ey R
espo
nse
GBI
Month of Overlap Period (t)
GBI in month t-1 Survey in month t
14
GBI vs. Monthly Close of Stock Price
Mean Coeff % p-val <.05
Constant -68.09 0.16
S&P* 0.10 1
GBI(t) -16.43 0.01
GBI(t-1) 30.81 0.99
Adj R-sq .46
15
Regressed brand’s stock price against GBI (controlling for market index). Across iterations…
Social Media Intelligence vs. Social Media Noise
• Account for differences in social media sentiment across venues and across sub-topics
• Potential to use social media for market research – Adjusted measure (GBI) can serve as lead indicator
• At a minimum, dashboard metrics should disaggregate and
separately report venue-specific and topic-specific metrics
16
Questions?
Why Do People Post? Opinion Formation versus Opinion
Expression
• Opinion formation
– Pre vs. post purchase – Customer satisfaction and word-
of-mouth
• Opinion expression
– Opinion dynamics – Opinion polls and voter turnout
Pre-Purchase Evaluation
Purchase Decision and Product Experience
Post-Purchase Evaluation
Incidence Decision
Evaluation Decision
Posted Product Ratings
EXPE
RIEN
CE M
ODE
L IN
CIDE
NCE
& E
VALU
ATIO
N M
ODE
LS
SELECTION EFFECT
ADJUSTMENT EFFECT
18
Perils of Selective Listening
• Index of over/under-estimation of messages mentioning specific product attributes
Application Compatibility
Documentation/Support Price Ease of Installation
Brand Reputation
Blog only -27.73 20.06 18.72 -31.39 13.43
Forum only 81.71 -21.20 -6.32 74.03 -31.99
Microblog only -57.37 6.02 -7.50 -46.90 22.00
Blog+forum 34.25 -3.30 4.54 28.32 -12.29
Blog+microblog -44.08 12.32 4.26 -39.94 18.16
Forum+microblog 14.26 -8.00 -6.89 15.38 -5.81
Blog+forum+microblog 2.38 -.06 .35 2.15 -0.36
19