Academically-Practical and Practically-Academic Social Commerce Learnings in Interactive Media
-
Upload
lift-summit -
Category
Business
-
view
792 -
download
0
description
Transcript of Academically-Practical and Practically-Academic Social Commerce Learnings in Interactive Media
Academically‐Practical and Practically‐AcademicLearnings in Interactive Media
Wharton Interactive Media Initiative
www.whartoninteractive.com
Professor Eric T. BradlowK.P. Chao Professor
Professor of Marketing, Statistics, and Education Vice‐Dean and Director, Wharton Doctoral ProgramsCo‐Director, Wharton Interactive Media Initiative
THERE IS NO GREAT DIVIDE!
Academics Practice
Academically‐Practical Practically‐Academic
HOW DO I KNOW WHAT ACADEMICS KNOW AND HOW DO I KNOW WHAT PRACTITIONERS CARE ABOUT?
• WIMI Corporate Partners o Travel and Listen!
• Matchmaking Webinarso Take Corporate Partner Business Problems and Present Them to
the Academic Community
• My Own Academic Research
• Academic Research Conferences and WIMI‐Funded Research
WIMI’S “LEARNING NETWORK”
Global network of research partners
W H A R T O N I N T E R A C T I V E M E D I A I N I T I AT I V E
Wharton Lab for Publishing Innovation
WIMIResearch,
Student Placement Interns,Partners
MATCHMAKING WEBINARS
+
=
+
NOBODY KNOWS HOW MUCH TO PAY THEM!
ADVERTISING ATTRIBUTION * NOT LAST CLICK
* NOT EQUALLY SPREAD
What is the #1 Problem Today for Internet Ad Publishers?
WHARTON INTERACTIVE MEDIA INITIATIVE
Display advertising on media sites
ORGANIC TACKLES AD ATTRIBUTION!
ORGANIC DEVELOPED AND MANAGED A COMPLETE DIGITAL MARKETING STRATEGY FOR THE CLIENT, A NEW CAR MANUFACTURER
Sponsored search
Shopping sites
Advertiser sites
User 3
User 2
User 1
Day 6
Day 20
Day 1
DATA
DIGITAL ADVERTISING “PATHS” FOR NEW CAR SHOPPERS (HYPOTHETICAL)
View AdEdmunds.com
View AdCNN.com
View AdCNN.com
Click‐through @ CNN.com
View AdCNN.com
Click‐through @ Google “Conversion” at advertiser site
View AdKBB.com
Page view at advertiser site
DATA
Display advertising impressions
• User• Date & time• Advertiser organization (i.e., brand)• Media buy name• Site where ad was displayed (28 sites)• User’s country, state & area code (based on IP)
For each activities at the advertisers site
(including conversions)
• User• Date and time• Type of activity
• “Conversion” or “Success” activities• Search inventory• Find a dealer• Build & price• Get a quote
• Other activities• User’s state & area code (based on IP)• Whether the conversion occurred in the
same session as a click‐through
Click‐throughs
• User• Date & time• Advertiser organization (i.e., brand• Media buy name• Site where ad was displayed• Ad id number
(no info on ad content)• User’s country & state code
(based on IP)
AVAILABLE FIELDS
KEY IS HAVING ALL THREE LINKED TOGETHER
What Is The # 1 ProblemToday In Search, From the Search
Firm’s Perspective?
WHARTON INTERACTIVE MEDIA INITIATIVE
WHAT TO SHOW WHEN SOMEONE SEARCHES?
Retail
EXPEDIA TAKES ON OPTIMAL SEARCH RESULTS
Corporate
Package
Media
Opaque
DATA ON 10,000K+ HOTEL SEARCHES CONDUCTED OCT 1-15, 2009
Travel Dates
Time/Date of Search
Number of Rooms
Number of Travelers
Free Text Associated with
Search Region/Distinct Keyword Assigned to that Text
FOR EACH SEARCH WE OBSERVE WHICH HOTELS WERE DISPLAYED
Number of Hotels that Meet Search Criteria
Hotels Displayed
Price Displayed for Each Hotel
Was the Price a Promo?
WE ALSO OBSERVE WHICH HOTELS WERE VIEWED AND PURCHASED
Which Hotels Got Click‐Throughs?(if any)
Which Hotels got “Book It” Click‐Throughs?
Which Hotels Were Purchased?
ERIC “THE WIZARD”: PREDICTING AND MONETIZING FUTURE BEHAVIOR
W H A R T O N I N T E R A C T I V E M E D I A I N I T I AT I V E
DESCRIPTION OF DATA
• Contains 23,000 users of hulu.com who registered during February 2009.o Take 10% random sample
• Tracking daily incidence of visiting to view videos for each of 120 days starting March 1, 2009.
• Summary Statistics of 90‐day in‐sample period:o Reach: 46% of people visit at least onceo Frequency: 4.3 visits on average, among those who visit o Streakiness: 446 total streaks of visits lasting 3 or more consecutive days (across all
people)
• Last 30 days are the holdout (out‐of‐sample) period used for model validation.
1 7
EXAMPLE 1: MAKING MONEY FOR HULU
1 8
ALIVE, THEN DEAD
ALIVE, THENCOLD
ALIVE OR “DEAD”
ALIVE OR COLD
WINNER ‐> DON’T PAY TO BRING BACK FROM THE “DEAD”
• A retailer (with catalogs, stores and a website) would like a tool to identify which consumers active and which ones have ended their relationship with the firm
• The retailer provided transaction history across three channels (web, store & catalog) for a random sample of 30,000 customers
• Using this data, researchers at WIMI are developing a model that can be used to:
o Identify ‘inactive’ customerso Forecast future saleso Plan capacityo Understand multi‐channel behavior
• Unlike many other forecasting approaches does not require any information about the consumer other than her purchase history
o Easily applied in many settings
MAKING MONEY FOR MECOX LANE
REMARKABLE PREDICTION ACCURACY
0
1000
2000
3000
4000
5000
6000
7000
8000
Cummulative Orders
Actual cummulative Forecast Cummulative
• The model is based on the simple idea that people buy at a steady rate until they become inactive
o But different people have different rates
• By using the data to estimate the rates at which people buy and become inactive, we build a model that can forecast orders into the future
• These models have proven accurate across many industries and contexts
* All results preliminary
INSIGHT INTO DIFFERENCES BETWEEN CHANNELS
0.0
100.0
200.0
300.0
400.0
500.0
600.0
700.0
800.0
900.0
1000.0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014
Proportion of Customers (Probability Density)
Daily Dropout Rate (!)
Beta Density for Dropout Rate
Overall Catalog (Method=1)
.com (method=8) Store (Method=M)
.com customers are less likely to drop out than others
Catalog customers are more likely to drop out than others
* All results preliminary
• Even though we never observe when a customer becomes inactive, the model gives us an estimate of the drop‐out rates
o Most .com customers have a very low drop‐out rate
o Most catalog customers have a much higher drop‐out rate
o Store shoppers vary widely in in their propensity to drop out
FORECASTING MULTI‐CHANNEL MEDIA CONSUMPTION DURING THE WORLD CUP
Wharton Interactive Media Initiative
Elea McDonnell FeitPengyuan WangEric T. Bradlow
Peter S. Fader
CONTEXT
ESPN OBJECTIVE FOR WIMI
Build a state‐of‐the‐art predictive model to
understand and project "multichannel"
consumption habits across digital properties
(Internet, Mobile & Streaming Video)
MULTI-CHANNEL TOURNAMENT FORECASTING
• The Wharton Interactive Media Initiative developed a state‐of‐the‐art predictive modeling method to understand and project multi‐channel consumption habits across media platforms (web, video and mobile).
• We tested this model using usage data for individual fans across three channels and were able to make accurate forecasts, measure the relationships between channels, and estimate the media attractiveness of individual teams.
6/4
6/5
6/6
6/7
6/8
6/9
6/10
6/11
6/12
6/13
6/14
6/15
6/16
6/17
6/18
6/19
6/20
6/21
6/22
6/23
6/24
6/25
6/26
6/27
6/28
6/29
6/30 7/1
7/2
7/3
7/4
7/5
7/6
7/7
7/8
7/9
7/10
7/11
Soccer Reach
(number of registered fans visiting daily)
Day
Soccer Reach for ESPN Digital Properties During World Cup
.comVideoMobile
US
US
US
US
FINDINGS
• Forecasting
o By summing up predictions for individual fans, we make accurate forecasts of overall reach for each channel.
• Multi‐channel behavior
o Fans are less likely to use ESPN.com on weekends, but Mobile usage is unaffected by weekends.
o Among those who use mobile and .com, the more a fan uses Mobile, the less he uses ESPN.com.
• Team strengths
o The method we have developed can be used to estimate the media attractiveness of individual teams.
0200400600800100012001400160018002000
6/4 6/11 6/18 6/25 7/2 7/9
Cumulative Frequency
(Total visits fo
r 100
0 registered
fans)
Cumulative Soccer Frequency for ESPN.com
during World Cup
0
20
40
60
80
100
6/4
6/6
6/8
6/10
6/12
6/14
6/16
6/18
6/20
6/22
6/24
6/26
6/28
6/30 7/2
7/4
7/6
7/8
7/10
Reach
(Num
ber o
f registered
fans visitin
g daily
per 1
,000
fans)
Daily Soccer Reach Forecast for ESPN.com
during World Cup
Predicted Actual
ACADEMICALLY PRACTICAL INTERACTIVE MEDIA 2009-2010
Social Networking: Jan 2009
Impact and Emergence of UGC: Dec 2009
Cross‐Platform Data: Dec 2010
Mobile Marketing: 2011
Use analytics to explore the relationship between brands
Text mine consumer postscompact sport old
Audi A6 67 345 56
Honda Civic 1384 539 245
Toyota Corolla 451 128 211
Mine Your Own BusinessMarket Structure Surveillance through Text Mining
Feldman, Goldenberg, Netzer
Customers are telling us things for “free”
Perceptual Map of US Car Makes
Is “classic” Marketing Research dead?
Does Chatter Really Matter? Dynamics of User-Generated Content and Stock Performance
Tirunillai and Tellis
What your customers are saying matters (if you own stock)
Short‐term
effect on
stock
returns
Long‐term
effect on
stock
returns
Chatter 3.8 4.8
Consumer Opinion
‐2.1 ‐3.6
Negative Chatter
‐2.9 ‐3.9
Negative Expressions
‐3.7 ‐4.7
“You can take UGC to the Bank”
Crowdsourcing New Product Ideas
Bayus
The value of the crowd is in the “crowd”
Daily: Feb 2007 – Feb 20097,100+ ideas
4,300+ ideators170 ideas implemented
Prior Experience Relationship to Future
Performance
# prior good ideas# prior reviewed ideas# prior ideas# prior comments
not significantnot significantnot significantnot significant
“The goal is for you, the customer,is to tell Dell what new products or
services you’d like to see Dell develop.”
Modeling Connectivity in Online Networks
• Social network data helps to improve predictions of behavior above and beyond just behavior
• More popular social networkers are also more active
• Online popularity is a more important correlate of online behavior than offline
Ansari, Koenigsberg & Stahl
Knowing a customers social graph helps predict their purchases
>
+ >
Econometric Modeling of Social Interactions
Hartmann
Consumers bring additional value through their community
Promote to Michael
Michael goes golfing
more
Michael’s friends golf
more
Direct Value65%
Indirect Value35%
Fraction of customer’s value that derives from
others in the group
Opinion Leadership and Social Contagion in New Product Diffusion
Target social influencers
Physician most often nominated by his peers as influential is targeted and is
persuaded to increase his/her prescription by 10 units
Iyengar, Van den Bulte, Valente
Influencers work, but slowly and “locally”
Across the board promotionEach physician is given an
additional detailer visit
vs.
But, free is free!
Popularity begetspopularity; but
how do you get it?
Pricing Digital Content
Iyengar, Abhishek, & Bradlow
Freemium works!
The Future: Data Minimization
www.whartoninteractive.com
37
LONG-STANDING IT CHALLENGE
38
TOO-MUCH-DATA PROBLEM
39
DATA PRIVACY ISSUES
SOLUTION: KEEP WHAT IS NEEDED, FIT WHAT IS THERE
• It is all about the data!o In many cases, practitioners have it – academics want it.o Scraping programs mean we can now all have it and in real‐time.
• Convergence of problems between academia and practice, in the interactive media space, has never been higher.o Advances still need to be made on scale of academic methods.
• Let’s look for the next great divide! It demonstrates an opportunity for further study.
S U M M A R Y
WHARTON INTERACTIVE MEDIA INITIATIVE
Eric T. [email protected]
www.whartoninteractive.com
W H A R T O N I N T E R A C T I V E M E D I A I N I T I AT I V E