discovery oriented social media research from descriptive to analytic (2011奢侈品panel分享)

Discovery-Oriented Social Media Research:

from Descriptive to Analytic

Deqiang ZouSchool of Management, Fudan UniversityNovember 2, 2011

© Copy rights reserved, Zou Deqiang

Why We Need to Know Research

• From a professor? After all, we can get Best practice from leading-edge business experts Latest observations and insights from marketing research or

consulting firms, and

• Professors always have some theories Unfortunately, I do not have any theories particularly for luxury

goods

• If no theories, maybe you can expand my horizon By showing me what I don’t know

2

3

GregMiller(2011),“SocialScientistsWadeintotheTweetStream,”Science,333(6051),1814‐1815.

Political Science


The Global Mood: Psychologist

4

Miller(2011)


Discovery Orientation

• This is not my job Transmit the knowledge

• I can share something with you A discovery oriented, critical, and generalizable mindset, so that

you’d be a Qualified consumer or even producer of social media research

• The key A transformation: from descriptive to analytic

5


Descriptive: Military Budget

6http://www.guardian.co.uk/news/datablog/2010/apr/01/information‐is‐beautiful‐military‐spending

Analytic: Who’s the Big Spender


Descriptive: Who has more soldiers?

8

Analytic: We need a ratio

11GroupM andCIC(2011),“TheVoiceofLuxury:SocialMediaandLuxuryBrandsinChina,”GroupMKnowledge‐ CICWhitePaperonLuxury,August.


So What?

• Buzz volume vs. Sales volume?• NSR vs. Brand attitude in the market place?• We need marketing research to support marketing

decision making If X, then Y Straightforward? No, marketing is losing its clout. We cannot be over-optimistic “Marketing has lost its seat at the (boardroom) table.” (from a

CMO)

12V.KumarandDenish Shah(2009),“ExpandingtheRoleofMarketing:FromCustomerEquitytoMarketCapitalization,”JournalofMarketing,73(6),119‐136.


The Cruel Reality

• The CMO is currently the most frequently fired C-level executive, with an average tenure of less than 24 months (Welch 2004)

• A research based on a multi-industry sample of 167 firms finds that the CMO presence in top management teams has almost no impact on firm performance (Nathand Mahajan 2008)

• There’s a perceived lack of marketing accountability, which has undermined marketing’s credibility, threatened marketing’s standing in the firm, and even threatened its existence as a distinct capability within the firm

13

KumarandShah(2009)


Establish the Accountable Relation

• Possible reasons lie in the failure of the marketer to accurately prove his or her worth and/or the inability to relate marketing performance to reliable financial metrics (Lehmann 2004; Rust, Lemon, and Zeithaml 2004)

• This entails relating marketing performance to a higher-level financial metric that is of concern to the CEO

• CIC and its partners have taken initiatives in attaching a monetary label to each online buzz

14

KumarandShah(2009)

15

R3andCIC(2010),InternetWordofMouthProventoHaveImpactonAutoSalesinChina.

What is the Causality?

R3andCIC(2010)


Causality: Temporal Sequence

Take the Teane for example, its positive IWOM led the way by one month for positive sales results R3andCIC(2010)


18

Confounding? (I)

R3andCIC(2010)


Confounding? (II)

19R3andCIC(2010)


20

Discovery: Three Components

•Y=f (x)

So what?

How?

Why?

21Nielsonwire (2011),HowSocialMediaImpactsBrandMarketing,Oct14.


Diagnose: One by One

• Why Differentiated effect of positive vs. negative WOM Other marketing efforts, e.g., pricing, channel, communication?

• So what Sales volume or incremental sales volume?

• How Clarify the causal relation, the underlying mechanism WOM can be the consequences of other marketing efforts Where is competition?

22

23

Price Elasticity• The refrigerated juice own- and cross-elasticities

a The percentage change in the sales of MinuteMaid with response to a 1% change in the price of Tropicana

DominiqueM.Hanssens,LeonardJ.Parsons,andRandallL.Schultz(2003),MarketResponseModels:EconometricandTimeSeriesAnalysis(2nd Ed.),Springer.

Groupon’s Influence on Reputation

24

http://www.technologyreview.com/blog/arxiv/27150/(Sep12,2011)

Source:http://arxiv.org/abs/1109.1530http://www.36kr.com/p/46496.html(Chinese)


Third party reports

Conceptual Framework of UGC and Stock Performance

GerardJ.Tellis andSeshadri Tirunillai (2011),“DoesChatterReallyMatter?ValidityandMeaningfulnessofUser‐GeneratedContent(UGC)forBrandEquity,”MarketingScholarForum(XI),Beijing:PekingUniversity.


UGC Metrics• Chatter (volume)• Ratings • Positive Chatter• Negative Chatter• Competitor metrics

(chatter, positive, negative)

Brand Performance• Abnormal Returns• Risk• Trading Volume

Summary of the Measures

Tellis andTirunillai (2011)


Effect of Chatter on Brand Performance

Returns Risk TradingVolume

Chatter +++ 0 +++NegativeChatter ‐‐‐ ++ ‐‐‐PositiveChatter 0 0 0

Ratings 0 0 0

Brand Performance

(Based on Granger Causality Test)

Cause



Effects of UGC On Brand Equity

Returns

ChatterImmediate 5%

Accumulated 15%

RatingsImmediate 1%

Accumulated 2%

Negative ChatterImmediate ‐3%

Accumulated ‐7%

Positive ChatterImmediate 1%

Accumulated 3%



Effects of UGC On Competitor Equity

• Any opportunities for new research? A missing link: online buzz the product market the financial

market

ReturnsChatter Immediate 5%Competitor Chatter

Immediate -1%Accumulated -2%

Competitor Negative Chatter

Immediate 2%Accumulated 3%



Online Consumer Review and Box Office Revenue

• The volume, but not the valence, of online user reviews has a positive impact on box office revenues (Liu 2006)

• The valence, but not the volume, of online user reviews has a positive impact on box office revenues (Chintagunta et al. 2010), exploiting the sequential release of movies across markets

30YongLiu(2006),“WordofMouthforMovies:ItsDynamicsandImpactonBoxOfficeRevenue,”JournalofMarketing,70(3),74‐89.

Pradeep K.Chintagunta,Shyam Gopinath,andSriram Venkataraman(2010),“TheEffectsofOnlineUserReviewsonMovieMox OfficePerformance:AccountingforSequentialRolloutandAggregationacrossLocalMarkets,”MarketingScience,29(5),944‐957.


The $ Value of the Star

31

Luca(2011)


It Pays!

• The impact of consumer reviews on the restaurant industry: a one-star increase in Yelp rating leads to a 5-9 percent increase

in revenue, this effect is driven by independent restaurants; ratings do not

affect restaurants with chain affiliation, and chain restaurants have declined in market share as Yelp

penetration has increased

32MichaelLuca(2011),“Reviews,Reputation,andRevenue:TheCaseofYelp.com,”HBSWorkingPaper,September.

35GroupM andCIC(2011)


Our Research Plan

• Relate monthly car sales in China to CIC data Buzz volume and valence at attribute level Incorporate competition effects

• Even more ambitious idea Relate OCJ sales data to natural language analysis of its

infomercial at attribute level

• Because analysis at attribute level is highly informative and insightful Analyze and visualize market structure by automatically eliciting

product attributes and brand’s relative positions from online customer reviews (Lee and Bradlow 2011)

36

37

LeeandBradlow (2011)

38ThomasY.LeeandEricT.Bradlow (2011),“AutomatedMarketingResearchUsingOnlineCustomerReviews,”JournalofMarketingResearch,48(5),881‐894.

Mapping the Market Using Customer Reviews

39Layconsumerwouldnotice:easy‐to‐usemenusandnavigationinterfaces,thenumberofpicturesavailable,andvideocapabilities

Technicallysophisticatedconsumers:low‐lightorISOcontrolsandlenscharacteristics(e.g.,namebrandopticssuchasZeiss)


Evolution of Market Structure

40



Market Structure by Cons and Pros

41


42CIC,新浪 (2011),“微博引领的中国社会化商业发展与变革,” CIC·新浪合作微博白皮书,10月.


We Care About, More

• Tweet diffusion model for Burberry 3D fashion show

43

GroupM andCIC(2011)


Descriptive Statistics

44

GroupM andCIC(2011),“TheVoiceofLuxury:SocialMediaandLuxuryBrandsinChina,”GroupM Knowledge‐ CICWhitePaperonLuxury,August.


Scale and Speed

• To profile the consequences of the tweet diffusion process Scale: how many people are infected? Speed: how fast are people infected?

• Infection? Yes, fashion is contagious. Its diffusion is just like epidemics

45


46

• In recent years, scientists have improved disease surveillance systems that enable public health officials to follow the emergence and spread of infectious diseases Most systems collect information from the entire population, such as

how many people have visited the doctor for flulike symptoms and how many flu tests have been submitted to the health department

In recent flu seasons, the lag time—the gap between a person showing symptoms of an illness and that data being available to epidemiologists—has decreased to as little as 1 week

• Although this improvement gives epidemiologists a better idea of how many people are ill, it doesn’t allow them to track an outbreak in real time or anticipate its spread

http://news.sciencemag.org/sciencenow/2010/09/social‐network‐predicts‐flu‐spre.html

Can the Diffusion Process Be Predictive?


47

Social Network Predicts Flu Spread

• Researchers who tracked flu symptoms in the friends of a group of college students during the 2009 H1N1 "swine flu" pandemic predicted the flu outbreak in the general college population with at least 2 weeks' advance notice Randomly chose 319 Harvard undergraduates, who then named

425 of their friends. Checked on the health of these 744 students between 1 Sep 2009, and 31 Dec 2009 using two different methods: a twice-weekly e-mail survey and the students' records at the campus health clinic

Students in the friend group showed signs of the flu between 14 and 69 days before the epidemic peaked in the control group of randomly selected undergraduates


48

Segmentation: Friendship Paradox

• Their predictions depend on a characteristic of social networks known as the friendship paradox, which states that your friends have more friends than you do Although it seems that on average your friends should have the

same number of friends as you do, a person named as a friend actually has more friends than you, because people named as friends tend to be more popular

They also tend to be better connected and more central to the social network

• Previous research showed that well-connected people in a network caught infectious diseases before those with fewer connections


Progression of flu contagion in the friendship network over time


50

Differences in Contagion

NicholasChristakis:Howsocialnetworkspredictepidemics(Sep2010),http://www.ted.com/talks/nicholas_christakis_how_social_networks_predict_epidemics.html


51

Is it Marketing Relevant?

• "What our method offers is the premise of predicting the future," Christakis says To apply the strategy to the general population, he notes, all

researchers would have to do is ask a randomly selected group of people to identify their friends and then track when these friends become ill. "Today, you can know where the epidemic will be in 2 weeks.“

• This lead time can give public health officials more time to develop an effective response to the outbreak

• In the context of social media We care about the “lead time” How to define “a fiend”?


The Long-term Downside of Overnight Success

• Professor Jonah Berger, and Gael Le Mens tracked the popularity of first names over 100 years in France and the United States The names that soar into popularity fastest, they discovered,

also tend to fall out of favor more quickly. "We often see products, ideas and behaviors catch on and

spread like wildfire. New high-tech gadgets or YouTube videos go from unknown to amazingly popular," says Berger. "But we know less about why once-popular things become unpopular."

52

BergerandLeMens (2009)


A few trajectories of first-name popularity

53JonahBergerandGaelLeMens (2009),“HowAdoptionSpeedAffectstheAbandonmentofCulturalTastes,”ProceedingsoftheNationalAcademyofSciences,106,8146‐8150.

54BillMarshandAliciaDeSantis (2009),“QuickArrivingFadsQuicktoFlameOut,”TheNewYorkTimes,May16.


100 Years of Names

• Names are also a good proxy for products and services that convey symbolic meaning about identity

• Being the same, yet different parents' attitudes toward naming their children reflect a

fundamental tension between an individual's desire to conformand fit in with others, while at the same time preserving a distinctive identity

• That’s how people buy luxury goods

55


Social Preference in Conspicuous Consumption

• Snobs

• Conformists

56WilfredAmaldoss andSanjayJain(2005),“PricingofConspicuousGoods:ACompetitiveAnalysisofSocialEffects,”JournalofMarketingResearch,42(February),30‐42.

1 1 1 1 1( , ) (1)e es s s sU z p v p t z

1 1 1 1 1( , ) (5)e ec c c cU z p v p t z


Settle the Conflicting Motives

• Conflicting motives for similarity, identity-signaling, and uniqueness can be resolved at different product levels People tend to choose options preferred by in-group members

on dimensions linked to their social identities (e.g., brands), and this is driven by desires to be associated with those groups

Higher needs for uniqueness lead people to make differentiating choices among group associated options (i.e., select less popular products from in-group associated brands)

• Evidence in social media? Effects with regard to their WOM behavior?

57

CindyChan,JonahBerger,LeafVanBoven (2011),Differentiatingthe“I”in“In‐Group”:HowIdentity‐SignalingandUniquenessMotivesCombinetoDriveConsumerChoice,JournalofConsumerResearch,ConditionallyAccepted


Two Types of WOM

• A psychosocial cost associated with positive WOM: positive WOM can decrease the uniqueness of one’s

possessions, which hurts high-uniqueness individuals

• High-uniqueness individuals are less willing to generate positive WOM for publicly consumed products that they own

• They are as willing to discuss product details• How to figure them out

In the context of social media?

58AmarCheema andAndrewM.Kaikati (2010),“TheEffectofNeedforUniquenessonWordOfMouth,”JournalofMarketingResearch,47(3),553‐563.


Two Types of Online WOM?

59GroupM andCIC(2011)

Conspicuous Consumption and Race

60

• Blacks and Hispanics devote larger shares of their expenditure bundles to visible goods (clothing, jewelry, and cars) than do comparable Whites Charles,Hurst,andRoussanov (2009)


Alternative Explanations

• Inherent ethnic difference • Conspicuous consumption is used as a costly indicator

of a household’s economic position Status signaling: individuals derive utility from status, which

depends on others’ beliefs about their income. Although income (or wealth) is not observed, visible consumption is

Visible consumption rise in own income, and fall in the income of the reference group— for each race group

An individual’s reference group is defined as persons of the individual’s race living in his state

• Reference group in social media

61Kerwin KofiCharles,ErikHurst,andNikolaiRoussanov (2009),“ConspicuousConsumptionandRace,”QuarterlyJournalofEconomics,124(2),425‐467.

Not everything that can be counted counts, and not

everything that counts can be counted

- Albert Einstein

Computers are useless. They can only give you answers.

- Pablo Picasso


64

Discovery: Review

•Y=f (x)

So what?

How?

Why?


65

YuvalAtsmon,Vinay Dixit,GlennLeibowitz,andCathyWu(2011),“UnderstandingChina’sGrowingLoveforLuxury,”McKinseyConsumer&ShopperInsights,McKinseyInsightsChina.

Return on Marketing


67

JonIwata(2011),“FromStretchedtoStrengthened:InsightsfromtheGlobalChiefMarketingOfficerStudy,”C‐SuiteStudies(CMO),IBM.


[email protected]

discovery oriented social media research from descriptive to analytic (2011奢侈品panel分享)

Business