Volume 5 of the Bazaarvoice Conversation Index

The
Conversation
Index

Volume 5

The Conversation Index
Volume 5

TheConversationIndex.com
Get your digital copy TheConversationIndex.co.uk
TheConversationIndex.de

#BVINDEX5

Table of Contents
What to expect........................................................................................................................................ 6
What we’ve found.................................................................................................................................... 9
Stock prices move with Twitter mentions...................................................................................................... 10
Twitter evolves from portal to destination..................................................................................................... 12
Brands get more tweets, but less of the conversation is about them................................................................. 16
Search interest doesn’t correlate to Twitter mentions, stock performance, or TV & radio coverage........................ 20
The bottom line? Social and “the real world” are becoming inextricable........................................................... 24
The methodology behind The Conversation Index Volume 5 / Contributors...................................................... 27
Contact us............................................................................................................................................... 28
About Bazaarvoice ................................................................................................................................... 31

5

THE CONVERSATION INDEX VOLUME 5

What to expect
We’re used to hearing social discussed as if it’s some kind of vast parallel universe, free of the cause and
effect of “the real world.” This view gets at least one thing right: the social universe is vast, and like our
physical universe, mostly unexplored. But where does social intersect with our everyday experiences,
in life and in business? How do social and the real world affect and reflect one another on a larger scale?
We’re starting to see patterns emerge that tell us that social refuses to contain itself; its effects are spilling
into the real world nearly everywhere we look: in politics, innovation, education, and of course, business
and the economy.

In this volume of The Conversation Index, we ride along the collision course of social data, traditional
media, and business performance, with an emphasis on Twitter. With over half a billion active users, and
an average of 340 million tweets per day, Twitter is like a social seismograph for the entire world. We
worked with enterprise social data provider Gnip to acquire 26 million tweets for this analysis. Every
tweet in this research mentioned at least one of 13 brands from the BrandZ™ Global 100 Brands list,
including Adidas, Clinique, Colgate, Gillette, Hugo Boss, Nike, Pampers, Pepsi, Ralph Lauren, Samsung,
Intel, Tesco, and Sony.

6

#BVINDEX5

We compare, contrast, and combine those 26 million tweets with over 8,000 TV and radio mentions, 17 months
of stock price data, more than a year and a half of Google search interest data, and 270,000 pieces of consumer-
generated content from online reviews—all for the same 13 brands.

This Index reveals some fascinating patterns and relationships at the intersections of all these data streams, as
well as a few places where we found—counterintuitively—no correlations at all.

Social data offers a critical new stream of insights for brands and the industry. The social ecosystem is broad;
conversations happen on brand sites and on social channels, across an ever-increasing set of devices. As social
content and social data expands, so does the scope of our knowledge about how it both influences and mirrors
activity across the digital and physical worlds. Sharing these insights with our clients and the industry is the
driving force behind The Conversation Index. We’re excited to share this edition with you.

Erin Nelson (@erinclaire)
Chief Marketing Officer, Bazaarvoice

7

#BVINDEX5

What we’ve found
Here’s what we uncovered:

• Twitter volume for brand mentions is highly correlated with stock price

• Twitter is becoming a destination, not just a portal

• Brand mentions in Twitter lag behind overall Twitter growth

• Search interest for brands doesn’t correlate to Twitter mentions, stock performance, or TV and radio coverage

9


Stock prices move more about brands as they perform well on the exchanges,
but quiet down a bit as their stocks fall.

with Twitter mentions Other analyses have shown social data to reflect economic
factors. Product reviews mention price more when consumer
confidence is low (a -.66 correlation).1 In February 2009,
It is becoming clear that both quantitative and qualitative
when consumer confidence hit its lowest point, mentions
analysis of social data can be useful for establishing
of price hit their highest point, accounting for 11.5% of all
relationships and in predicting real-world events. Stock
US reviews. When we map these price references to the
performance for the brands in this analysis increased and
Dow Jones Industrial Average, an even stronger negative
decreased in line with the fluctuating volume of tweets about
correlation (-.68) is revealed. Price mentions fall as the Dow
these brands. This is a remarkably high positive correlation
average rises, and they rise as the Dow falls. And a 2010 study
of .91, meaning that high Twitter volume tends to coincide
by academic researchers at Indiana University and University
with high closing price, and vice versa. The same things that
of Manchester found that measuring select dimensions of
make stocks move upward tend to make social chatter spike
“Twitter mood” can be 86.7% accurate in “predicting the up
(such as well-received product announcements and high-
and down changes in the closing values” of the Dow Jones
profile executive hires). But there’s a piece of this finding that’s
Industrial Average.2 As more of these predictive relationships
counterintuitive. Shouldn’t the same factors that send a stock
are discovered and subject to rigorous testing, social data
downward (such as a lawsuit, product flop, or poor earnings
will be incorporated into things like demand forecasting and
call) be just as likely to trigger a bump in conversations about
product release timing.
the brand? Apparently not. It seems that Twitter users buzz

1
The Conversation Index Volume 1.
2
http://www.relevantdata.com/pdfs/IUStudy.pdf

10

#BVINDEX5

CLOSING PRICE CORRELATES TO TWITTER BRAND MENTIONS
200 200
190 180.2
190
180 173.7 180
stock performance and Twitter mentions.
170 170
159.2

AVERAGE BRAND MENTIONS ON TWITTER (in thousands)
155.5 156.6
160 160
150 142.6 150
135.4 136.9
140 134.2 132.9 154.3 140
130.9 130.9 131.4 129.9 129.9 131.4
AVERAGE CLOSING PRICE

125.6 144.8
130 121.1 130
133.7
120 120
124.5
110 110
100 100
107.1
90 99.7 90
80 80
88.4
70 80.8 70
73.4 74.8
60 68.2
60
68.1
50 62.4 62.2 60.9 50
40 51.0 40
45.8
44.7
30 30
20 20
10 10

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN
2011 2012

11


Twitter evolves from
portal to destination
As Twitter grows, it is becoming a destination rather than a
portal or midpoint. People are increasingly going to Twitter
for the experience it delivers — conversation and timely
information — not as an intermediate step between them and
what they’re really after. And they’re staying longer. But Twitter
is used much differently than other social and web channels,
and its data should be used differently, too.

For example, we compared online search terms that include
“Adidas” to mentions of “Adidas” on Twitter. Since search takes
place when someone is looking for specific information, the
terms that appear during a search vary greatly from other types
of content. Top “Adidas” search terms tend to be at the level
of product lines and categories, and of the top 20 searches,
only three are for specific products. The top 20 search terms
associated with “Adidas” also included the names of two
competing brands, indicating that comparison shopping
begins in the search box.

12

#BVINDEX5

Twitter users tend to reveal their personal interactions with include product suggestions, and a fifth of all four-star reviews
or in relation to brands. When we dig into what people on provide this type of feedback.3
Twitter are saying about Adidas, they mention what “new”
Use search patterns to optimize your brand’s search
Adidas products they’re wearing “today,” and use words such
engine marketing and optimization strategies.
as “my” and “I.” Some of the top Twitter terms associated
with Adidas reference specific campaigns, like the hashtag And use the words you find in tweets about your brand when
“#branch309adidasday.” Businesses interested in doing you tweet about your brand. Optimize your marketing copy by
their own text analysis on tweets should think of it as a way using the language of your top reviewers—or better yet, quote
to roughly gauge consumer response to news, events, them. When you reflect what consumers say online in your
campaigns, and as a method of identifying enthusiastic own social efforts, you’ll create content that’s more shareable.
customer advocates. But they will have to develop repeatable Quantify this over time by testing optimized tweets, UGC-
methods of separating relevant, authentic tweets from the rich copy, and SEM strategies derived from the actual terms
abundance of noise. people use when searching for your brand and products,
and compare each to the non-optimized, marketing-derived
In contrast to search and Twitter language, reviews tend to
alternative.
focus on specific product qualities (“light,” “looks,” “fits,”
“comfortable”), adjectives (“awesome,” “great,” “different,” Tweets that mention brands are using fewer links over time.
“perfect”), and other expressions of sentiment (“love,” In the last half of 2010, 68% of tweets that mentioned brands
“disappointed,” “happy,” “recommend”). Since every single also had links in them. In all of 2011, the number dropped
review is tied to a specific product, they are a rich social data to 55%. In the first half of 2012, the number drops further to
source for product-level insights. In fact, 12% of all reviews 51%, signaling a clear downturn in link usage. This means

3
The Conversation Index Volume 3.

13


TIME ON TWITTER AND PAGES PER VISIT ARE GROWING
780 20
760 Time on site (in seconds) 19
740 Pages per visit 18
720 17
700 16
15.1 15.2
680 14.1 15

714
684
660 14
12.9
TIME ON SITE (in seconds)

666
640 13
12.0

PAGES PER VISIT
620 12
10.8
600 10.3 10.4 9.9 11
10.1

602
9.9 9.7
580 10

583

583
578

579
593
560 8.4 8.4 9
8.1 8.1 8.0
540 7.7 7.6 8

551
7.4 7.3 7.4

566

544
6.9

563
520 6.6 6.5 7

528
500 6
480 5
486
460 4
467
458

457
456

440 3
448

434

420 2
428
425

423
421

400 1
G

P

T

OV

C

N

B

R

PR

AY

N

L

G

P

T

OV

C

N

B

R

PR

AY

N

L

G
JU

JU
DE

DE
SE

A

SE

A
AU

OC

AU

OC

AU
FE

FE
JA

JU

JA

JU
A

A
M

M
N

M

N

M
2010 2011 2012
Source: COMPETE.COM

14

#BVINDEX5

that content about brands on Twitter is becoming increasingly
conversational, and less transactional. Users are talking about
brands instead of just pointing to what they bought or want to
buy with a link to external sites.

External data confirms this. Twitter users are spending more time
on Twitter, and visiting many more pages within Twitter.com
while they’re there. According to data from Compete.com, from
2010 to 2011, there was a 19.8% increase in average time on site;
in 2011 to 2012, so far, there is a 19.7% increase in time on site.
And average pages per visit decreased 9% from 2010 to 2011,
but increased by an incredible 58.7% from 2011 to 2012.

Pages per visit
increased by an
incredible 58.7%
from 2011 to 2012

15


Brands get more retweets. So far in 2012, 22% of all brand mentions on Twitter
have been retweets, and only 78% of brand mentions

tweets, but less of are original.

the conversation is There’s good news and bad news for brands in this data.
The increase of brand mentions overall means there is more
about them data to learn your customers’ thoughts about you, but as the
retweet analysis shows, that data is increasingly redundant.
Retweets are becoming a bigger part of the Twitter brand
The volume of tweets per day has grown 143% from 2011
story, but a retweet is a weaker social signal than an original
to 2012; however, mentions of brands on Twitter have only
tweet from, say, an advocate or detractor. Retweets also
grown 113% in the same period. To maintain and improve
contain less original data, and may not represent the users
Twitter share of conversation, brands should analyze their data
behind them as much as a wholly original tweet from the same
to find which tweets are generating positive conversation
user. Our research also found that some of the most retweeted
about them, emulate these tweets, and continuously optimize
content is the work of automated bots (nonhuman scripts)
and add fresh content.
and spammers that have set up networks of auto-retweeting
Original tweets about brands are declining over time, accounts to spread their inauthentic messages across the
as retweeted brand mentions are rising. social web as quickly as possible before Twitter shuts them
In other words, more and more content is simply repeated down. Altogether, this means that businesses need to apply
verbatim or with little alteration from the original source. In more scrutiny to Twitter data. Perform spot checks, weight and
2010, 85% of brand mentions on Twitter were original, and filter your metrics to place less emphasis on retweets about
15% were retweets. In 2011, 18% of brand mentions were your brand if you find they are far more noise than signal.

16

AGGREGATE OF BRAND MENTIONS (in thousands)

2500

500
1000
1500
2000
JU
L 537,332
AU
G 539,864
SE
P 536,894
OC

2010
T 615,218
second half of 2010.

N
OV 638,983
DE
C 587,469
JA
N 681,455
FE
down from 82% in 2011 and 85% in the

B 695,041
78% original tweets in 2012 (as of June),

M
A 784,875
R
A
PR 879,132
M
AY 1,031,875
JU
N 919,238
JU
L 897,434
2011 AU
G 967,993
SE
P 1,039,532
OC
T 1,130,981
N
OV 1,067,332
DE
C 1,247,513
JA
N 1,372,390
FE
B 1,507,232
M
A 1,724,314
R
A
PR 1,829,344
M
AY 1,977,783
2012
BRAND MENTIONS VOLUME GROWING; ORIGINAL TWEETS DECLINING

JU
N 2,229,620
10%
70%

30%

20%
50%
60%
90%

80%

40%
100%

ORIGINAL TWEETS (i.e., not retweets)
17
#BVINDEX5

18
PERCENTAGE OF TWEETS WITH LINKS

10
20
30
40
50
60
70
80
90
100
JU
L 537,332
AU
G 539,864
SE
P 536,894

2010
OC
T 615,218
N

OV
638,983
DE
C 587,469
JA
N 681,455
(as of July), down from 55% in 2011

FE
51% of tweets contain links in 2012

and 68% in the second half of 2010.

B 695,041
M
A
R 784,875
A
PR
879 , 132
M
AY
1,031,875
JU
N 919,238
JU
L

2011
897,434
AU
G 967,993
SE
P 1,039,532
OC
T 1,130,981
N
OV
1,067,332
DE
C 1,247,513
JA
N 1,372,390
FE
B 1,507,232
M
A
R 1,724,314
A
PR
1,829,344
M
2012

AY
1,977,783
JU
N 2,229,620
PERCENTAGE OF BRAND MENTIONS CONTAINING LINKS IS DECLINING

JU
L 994,291
500
1500

1000
2000
2500

AGGREGATE OF BRAND MENTIONS (in thousands)

#BVINDEX5

The increase in retweets also illustrates that news travels faster network (and, as this Index shows, beyond) by creating
than ever before—and that a single piece of content can have consistently retweetable content—they are the greatest
major consequences for the companies involved. In fact, distributors of social currency. Reach them, highlight, and
many of the most retweeted messages about brands in our promote them if they are advocates, and address their
analysis were highly negative in sentiment, and concerned concerns if they are detractors. Determine whether they are
things like scandals, lawsuits, and negative press coverage. influential in other channels: Are they a top reviewer as well?
Now is the time to prepare social crisis communications plans Give them exclusive access: insider news, early product
if you haven’t already. testing, event invitations, and the like. Make them feel like
a part of your brand instead of a spectator, and in all cases,
It’s also important for brands to get to know the real people
locate them as soon as possible.
that are creating the ripple effect for their brand across the

19


Search interest doesn’t
correlate to Twitter
mentions, stock
performance, or TV and Unpaid
radio coverage coverage
doesn’t
While it may seem that people tweet what’s top of mind, they’re
not tweeting about what they’re searching for. While we saw
drive much
this across the board, we’ll use Clinique as an example. Twitter
mentions for “Clinique” spike in April 2011, August 2011, and
search activity
March through June 2012. During these Twitter peaks, however,
we saw either no correlation with search interest or a decline in
search interest (search interest is Google’s normalized indicator
of “the likelihood of a random user to search for a particular
search term” on a 0-100 scale). When we compared the stock
performance of the brands in this analysis to search interest for
the same period in time, we found no correlation.

20

NETWORKS GROWING FOR USERS THAT MENTION BRANDS

FOLLOWER GROWTH FOR USERS THAT MENTION BRANDS
JUN 1566

MAY 1309

APR 1287
2012

MAR 1161
FEB 1299
JAN 1183
DEC 1041
NOV 1087
OCT 1155
SEP 1177
AUG 1194
JUL 1067
2011

JUN 1067
MAY 1208
APR 1130
MAR 993
FEB 1097
JAN 995
Average Twitter followers per user
0

0

0

0

0

0

0

0

0

00

00

00

00

00

00

00

00

00

00

00
10

20

30

40

50

60

70

80

90

10

11

12

13

14

15

16

17

18

19

20
22

#BVINDEX5

We analyzed 8,000 brand mentions in closed captioning
data from television broadcasts (most TV ads are not closed-
Brand captioned, so ad mentions are not reflected in this data), and

advocates
radio transcripts (this data does include ads) to determine
whether brands being mentioned in traditional media saw a

and detractors corresponding bump in search interest. Surprisingly, they do
not. This suggests that unpaid coverage (news pieces, etc.)

have wider doesn’t drive much search activity, but findings from a study by
Efficient Frontier show television ad campaigns correlating to

audiences in a 60%-80% bump in brand-name search during the life of the
campaign.4 So, while unpaid coverage in traditional media

2012 may be a great awareness mechanism, it’s not driving the
consumer search behavior many businesses are craving.
For that, television ads still seem to do the trick.

4
http://econsultancy.com/us/blog/7731-how-can-marketers-use-offline-ads-to-drive-people-online

23


The bottom line? Social Convergence is a new concept for many companies, but it’s
actually nothing new in practice. In fact, it’s the “place” we’ve

and “the real world” are called home since 2005. Reviews, Q&A, and stories are all
forms of earned social content that live on owned digital real
becoming inextricable estate. And while we were helping clients across the globe
integrate owned and earned, Twitter launched, Facebook
opened to the public, and search became more and more
The borders between “social” and “the real world” are social. Channels blossomed, and are now converging. Data
difficult to pinpoint, but they’re being redrawn in some places, exploded in volume and then fragmented, and is now coming
and eliminated in others. Consider this: Just a few years ago, together again. Convergence will soon cease to be the
the terms “in-store” and “online” gave us a clear, differentiated exception, and will become the rule, just as product reviews
way to talk about channels. But as channels converged, and on company websites were once the exception.
consumers began to use the mobile web while in the physical Social’s connection to the world around us is has been
aisles, the terms no longer accurately described the way established in some areas, cannot be found in others, and has
that people actually shop. The same is true of social—and yet to be discovered or quantified in most. But it’s far better
everything it touches. for businesses to look for it everywhere and find it only in
some areas, than for them to stumble over it where they least
expected it.

24


26

#BVINDEX5

The methodology behind
The Conversation Index Volume 5
Volume 5 is based on an analysis of social content and other data surrounding 13 brands appearing on the BrandZ™
Top 100 Global Brands list, which ranks the “most valuable global brands” of 2012. The brands analyzed are Adidas,
Clinique, Colgate, Gillette, Hugo Boss, Nike, Pampers, Pepsi, Ralph Lauren, Samsung, Intel, Tesco, and Sony.

The data includes 26,000,000 tweets, over 8,000 TV and radio mentions, 17 months of stock price data from relevant
exchanges, more than a year and a half of Google search data, and 270,000 pieces of authentic user-generated
content from online reviews across the vast Bazaarvoice network.

Contributors
Column Five Media created the visualizations for The Conversation Index Vol. 5.

columnfivemedia.com

27


Contact us
Contact us to see how we help brands gain invaluable consumer and product
insights by putting consumer conversations at the heart of their organizations.

United States: (866) 522-9227 Germany: +49.89.24218508
bazaarvoice.com bazaarvoice.de

United Kingdom: +44 (0) 208.080.1100 Netherlands: +31.20.301.2169
bazaarvoice.co.uk
Australia / Asia-Pacific: +61.2.9362.2200
France: +33 1 56 60 54 45
bazaarvoice.fr
Sweden: +46.8.463.1083

San Francisco: (866) 345-1461

28

#BVINDEX5

About Bazaarvoice
Bazaarvoice brings the voice of customers to the center of business strategy, transforming business performance for nearly
2,000 clients globally, including over half of the Internet Retailer 500 list of the world’s largest retailers, over
20 percent of the Fortune 500, and over one-third of the Fortune 100 brands. Bazaarvoice social software helps clients
like Best Buy, Costco, Dell, Macy’s, P&G, Panasonic, QVC, Travelocity, and USAA create social communities on their brand
websites and Facebook pages where customers can engage in conversations. These conversations can be syndicated
across Bazaarvoice’s global network of client websites and mobile devices, making the user-generated content that digital
consumers trust accessible at multiple points of purchase. Through Bazaarvoice, manufacturers can also connect directly
with consumers on retail sites to answer questions and respond to reviews about their products. The social data derived
from online word of mouth translates into actionable insights that improve marketing, sales, customer service, and product
development. Headquartered in Austin, Texas, Bazaarvoice has offices in Amsterdam, London, Munich, New York, Paris,
San Francisco, Stockholm, and Sydney.

For more information, visit www.bazaarvoice.com, read the blog at bazaarvoice.com/blog, and follow on Twitter at
twitter.com/bazaarvoice.

#BVINDEX5
TheConversationIndex.com
TheConversationIndex.co.uk
TheConversationIndex.de

31

Volume 5 of the Bazaarvoice Conversation Index

More Related Content

Similar to Volume 5 of the Bazaarvoice Conversation Index (20)

More from Brett Hurt (8)

Volume 5 of the Bazaarvoice Conversation Index