1. Word of Mouthâs
Role in Driving Sales
Greg Pharo
Director, Market
Research & Analysis
AT&T Matt Sato
Manager
Accenture
Brad Fay
COO
Keller Fay Group
June 13, 2011
ARF AM 6.0, New York, NY
2. Spending on WOM Rising Fast
âWord of Mouth Marketingâ and âSocial
Mediaâ Are Among the Most Exciting New
Tools in the Arsenal of Marketers Today
$3,043
$2,572
$2,204
$1,918
$1,701
$1,543
$1,351
$981
$722
$487
$313
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
WOM Marketing Spending WOM Marketing Forecast
Source: PQ Media
3. Does Word of Mouth Drive Sales?
Questions Remain on Word of Mouthâs Role
in Generating Sales
ďˇ Does word of mouth directly influence sales
volume, and to what extent?
ďˇ Where does word of mouth fit into the âowned-
earned-paidâ media model?
ďˇ Is it really a metric of interest to companies?
#ARFAM6
4. Background - AT&T Marketing ROI
ďˇ AT&T is one of the nationâs largest advertisers
ďˇ Well-developed Marketing ROI program
ďˇ Uses Market Mix Modeling to optimize DMA-
deployment of media
â Partnered with Accenture and Mediaedge to develop
advanced analytics capabilities for market mix
optimization
ďˇ AT&T also tracks weekly and monthly brand
awareness, attitudes, and usage with a multitude
of market research studies
#ARFAM6
5. Problem - âMetrics Clutterâ
ďˇ AT&Tâs tracking studies collect a constellation of
market metrics:
â Brand perceptions
â Usage
â Customer satisfaction
â Literally hundreds of data series
ďˇ Management wanted to know which metrics â in
addition to media - are most impactful on Mobility
sales (i.e., âGross Addsâ) and on disconnects
(i.e., âchurnâ)
#ARFAM6
6. Methods - Create a Purchase
Funnel Model
ďˇ AT&T and Accenture
created both a Purchase
Funnel model which
identifies which metrics
are the most significant
influencer Gross Adds
ďˇ The model also shows
what other upstream
metrics drive these
key metrics
#ARFAM6
7. Methods - Using a Two-step
Process to Identify Key Metrics
ďˇ Analytical techniques are used to winnow the
myriad of earned media metrics
â Highly-related metrics were grouped together using a
cluster analysis
â A short-list of metrics that are most correlated with their
group are selected
ďˇ These representative metrics are then input into
a separate model
â Reduces the burden of incorporating potentially hundreds
of metrics
â Ensures the earned media impact is not âdilutedâ
by having related metrics in the same model
#ARFAM6
8. Methods - SEM Modeling
ďˇ Traditional regressions ďˇ The SEM structure, used here,
assume no interactions allows for interaction among
among sales drivers sales drivers
Brand Health Brand Health
Gross Gross
Paid Media Paid Media
Adds Adds
Word of Mouth Word of Mouth
#ARFAM6
9. Methods -
Measuring ALL Word of Mouth
ďˇ Keller Fay Groupâs TalkTrackÂŽ, a
national syndicated program measuring
WOM in all forms Mode of Conversations
â Over 3 in 4 conversations occur face-to-face Across All Categories
ďˇ The study involves 36,000 online
consumers surveyed annually,
Other
â 100 every day 2% Face-
â Yielding about 1,000 weekly mentions of to-Face
brands; 350,000 per year 77%
Online
ďˇ Respondents are representative of the 6%
US population aged 13 to 69
Phone
â use a diary to keep track of their brand
15%
conversations, then complete an online
survey to gather detailed information about
these conversations
â Quotas/weights by age, gender, education,
race, etc.
#ARFAM6
10. Finding - WOM Is a Major
Driver of Sales
ďˇ The number of positive WOM âmentionsâ in TalkTrackÂŽ
proved to be one of the more powerful metrics directly
influencing âGross Addsâ (sales)
ďˇ Unaided Advertising Awareness, a top-of-funnel metric,
was also a strong driver of Gross Adds
ďˇ In turn, the Structural Equation Model identified which
metrics influence Word of Mouth and Unaided
Advertising Awareness
ďˇ Paid media drivers are also included, as they directly
impact Gross Adds, Word of Mouth and brand
health metrics
#ARFAM6
11. Unaided Ad Awareness and WOM Are Two
Strong Direct Influencers of Gross Adds
Unaided Ad
Awareness
Word of Mouth-
Positive Mentions
Device Gross Adds
perception #1
(non-customers)
Strength of Network
Relationship perception #1
(non-customers)
Strong
Moderate
Weak Provider
Consideration
#ARFAM6
12. The Model Also Identified Attitudinal
Metrics Which Influenced Word of Mouth
Customer Service
Perception #1
Network Network Willingness to
perception #2 Perception #3 Recommend
Word of Mouth-
Positive Mentions
Strength of
Relationship
Strong
Moderate
Weak Gross Adds
#ARFAM6
13. Word of Mouth Data Was âCleanâ Enough
to Model, in Contrast to Online âBuzzâ Data
ďˇ Word of Mouth variables were easily incorporated
into the model
ďˇ In contrast, online âbuzzâ data proved difficult to
incorporate into models
â Computer-scored online buzz sentiment data did not prove to
be as accurate as hoped
â Online buzz may not always include all relevant online sites
â WOM captures a broader spectrum of discussions; fewer than
10% of conversations are online
#ARFAM6
14. Next, Word of Mouth Was Trialed in
Traditional Market Mix Models
ďˇ AT&T next introduced Word of Mouth variables into
traditional Market Mix Models
â AT&T constructed market mix models for itself and
key competitors
â Each model uses Gross Adds as dependent variable
â Media, pricing, product innovation, messaging performance,
competitive, other relevant marketing/environmental factors
incorporated as independent variables
â Modeling Approach: Multiple regression analysis
ďˇ Word of Mouth proved to be a powerful and
statistically significant sales driver in Mix Models
â Word of Mouth explained 10%+ of sales volume
â Paid Media remains #1 sales driver, driving ~30% of sales â but
WOM is one of the top influencers of Gross Adds
#ARFAM6
15. AT&T Conclusions
ďˇ Word of Mouth is an impactful, relevant variable
for influencing sales in the Wireless category
ďˇ WOM metrics belong on a CMO dashboard as a key
performance indicator
#ARFAM6
16. AT&T Next Steps
ďˇ Leverage Word of Mouth data in other analytics
projects, including tactical campaign analysis
ďˇ Deeper learning on paid media/WOM interaction
ďˇ Making it actionable: influencing conversations
ďˇ Work with research vendors to improve quality of
online buzz data
#ARFAM6
17. Keller Fay Observations
ďˇ AT&T analysis provides strong evidence that
âconversationâ should be a marketing objective
â Today, about half of WOM is influenced by marketing, including
20% by paid advertising
â These numbers ought to grow as marketers adopt word of
mouth as an objective
ďˇ Ways to Stimulate WOM
â Messages should be âtalkworthyâ and easy to share
ďˇ Think about providing âtriggersâ
â Targeting: Aim for consumers with larger social networks
ďˇ Seek out âinfluencersâ
â Channels: Favor those that facilitate conversations
ďˇ Not just âsocial mediaâ, but any media that reaches people in a
social context
ďˇ Pay-off: Conversation, advocacy, SALES
#ARFAM6