SLIDE 1
2
Mobile Ad Effectiveness: Hyper-Contextual Targeting with Crowdednes - - PowerPoint PPT Presentation
Mobile Ad Effectiveness: Hyper-Contextual Targeting with Crowdednes - - PowerPoint PPT Presentation
Mobile Ad Effectiveness: Hyper-Contextual Targeting with Crowdednes 2 Mobile Targeting Motivation Ad spending: $100B by 2018 Key: reach consumers when and where most receptive 3 eMarketer 2014 Mobile Technology Portability =
SLIDE 2
SLIDE 3
Mobile Targeting Motivation
- Ad spending: $100B by 2018
- Key: reach consumers when and where most receptive
eMarketer 2014
3
SLIDE 4
Mobile Technology
- Portability = Real-time Targeting
- GPS, Wi-Fi, Bluetooth, iBeacon = Geo-Targeting
4
SLIDE 5
Mobile Targeting with Crowdedness
- Mobile technology can gauge crowdedness on-the-go
5
SLIDE 6
Research Objective
(1) How does crowdedness affect consumer response to mobile targeting? (2) What drives the results?
6
SLIDE 7
Research Design
- Ideal test of crowding effects:
- randomize crowdedness
- Our test:
- field data measuring crowdedness with mobile
technology
12
SLIDE 8
SLIDE 9
Overview of Results
- Crowding positively affects mobile ad purchase
- Crowding invades space so people turn inwards
- Results opposite of crowding literature
- Crowding in retail stores decreases purchases
- May be a different manifestation of avoidance
10
Harrel et al. 1980; Zhang et al. 2014
SLIDE 10
Overview of Results
- Paradox of crowded environment
- Noise distracts consumer attention to ads
- But, crowding boosts attention to signal of mobile ads
11
Bart et al. 2014; Ghose and Han 2014
SLIDE 11
Prior Research
11
SLIDE 12
Mobile Research
- Mobile internet search behavior
- Coupon redemption rates
- Time and location
(*my forthcoming Management Science paper)
- Geographic mobility
Ghose et al. 2013; Molitor et al. 2014; Luo et al. 2014; Ghose and Han 2011
12
SLIDE 13
Mobile Research
- In-store mobile promotions
- Product characteristics
- Cross-platform synergies
- Environmental factors
Hui et al. 2013; Bart et al. 2014; Ghose et al. 2014; Molitor et al. 2013
13
SLIDE 14
- Disease and juvenile delinquency
- Stress, frustration, hostility
- Felt loss of control
Crowdedness Research
Schmitt 1966; Collette and Webb 1976; Zimbardo 1969
14
SLIDE 15
- Avoidance behaviors
- Threatened sense of uniqueness
- Risk aversion
Harrell et al. 1980; Xu et al. 2012; Meang et al. 2013
Crowdedness Research
15
SLIDE 16
Weekend Day Business Weekday Exogenous Crowding
Field Data (Quasi-field experiments)
16
SLIDE 17
Measuring Crowdedness
- passengers/m²: Subway mobile users connect to subway-
specific cellular line
17
SLIDE 18
- Targeted subway population: 2 million commuters
- Sample size: pushed to 10,360 mobiles
- Weekday and weekend
Parts 1 & 2
18
SLIDE 19
Mobile Message
- 20 Minute Expiration
- Promotional Discount
19
SLIDE 20
(1) Peak hours vs. non-peak hours of crowdedness
- 5 times (7:30-8:30, 10-12, 14-16, 17:30-18:30, 21-22 hrs)
- Subway station and direction
(2) Weekdays and weekends
Self-Selection Threats
20
SLIDE 21
(3) Randomization
- Excluded users who had the service or received the SMS already
- Randomized remaining users and pushed SMS.
(4) Personal mobile usage habits
– ARPU – MOU – SMS – GPRS
Self-Selection Threats (cont’d)
21
SLIDE 22
- Same-train-same-time subsample analysis
Additional Self-Selection Approaches
22
SLIDE 23
- Propensity score matching
Additional Self-Selection Approaches
23
2.5 5.0 7.5 10.0 12.5 15.0 17.5
Percent
N 391 Mean 0.39 Median 0.38 Mode . Normal Kernel Normal
- 0.06
0.06 0.12 0.18 0.24 0.30 0.36 0.42 0.48 0.54 0.60 0.66 0.72 0.78 0.84 0.90 0.96 2.5 5.0 7.5 10.0 12.5 15.0 17.5
Percent
N 391 Mean 0.39 Median 0.38 Mode . Normal Kernel Normal 1
ps
Traffic
Treatment Control
SLIDE 24
2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 0.91 1.96 3.05 4.02 4.97
Purchase Rate (%) Crowdedness as Passengers/m²
Effect of Crowdedness
24
16% 49.5%
SLIDE 25
Endogeneity Threat
- Identification with street closures
20
SLIDE 26
Street Closure Crowdedness
1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5%
2.11 2.72 3.05 3.54 3.99 4.13
Purchase Rate Crowdedness as Passengers/m²
22
SLIDE 27
Parameter Model 1 Model 2 Model 3 Model 4 Crowdedness X Street Closures .492** (.187) Crowdedness .126** (.041) .114** (.042) Street Closures
- .120
(.117)
- .142
(.177)
- 1.887
(1.057) Ln(ARPU) .301** (.118) .308** (.119) .308** (.119) .306** (.119) Ln(MOU)
- .043
(.065)
- .043
(.065)
- .044
(.065)
- .044
(.065) Ln(SMS) .014 (.069) .014 (.069) .015 (.069) .013 (.069) Ln(GPRS)
- .001
(.024)
- .001
(.023)
- .001
(.023)
- .001
(.023) Day(weekday) Effects Yes Yes Yes Yes Train (time cycle) Effects Yes Yes Yes Yes Observations 11,960 11,960 11,960 11,960
Main Evidence for Crowdedness Effect
SLIDE 28
Endogeneity Threat
24
Identification with unanticipated train delays
SLIDE 29
4.0% 4.5% 5.0% 5.5% 6.0% 6.5% 7.0% 7.5%
3.04 3.14 3.63 4.71
Train Delay Crowdedness
Crowdedness as Passengers/m² Purchase Rate
25
SLIDE 30
Lower Threshold
Subsample with Low Crowdedness (under 2 passengers/m2) Parameter Model 1 Crowdedness
- .084
(.270) Mobile Behaviors Yes Day(weekday) Effects Yes Train (time cycle) Effects Yes Observations 2,886
30
SLIDE 31
Upper Threshold
32
SLIDE 32
More Evidence with Field Surveys
Participants: 300 Purchasers & non-purchasers Survey Response: 240 of 300 mobile users = 80%.
33
Immersion Involvement Crowdedness
.465*** 1.375***
Purchase
.152*
Mobile Immersion
SLIDE 33
THANK YOU!
Xueming.Luo@temple.edu
33