Laura Ludeña
Head of Market Insights Google Spain
Nuevos usos de la estadística en la sociedad del conocimiento y de la red
Eustat. Donostia Julio 2013
Google Confidential and Proprietary
1
Agenda
1.
Uso de la estadística para conocer el mundo
digital
2.
Uso de la estadística para mejorar los
productos publicitarios
3.
Uso de la estadística para medir la efectividad
de los productos publicitarios
Google Confidential and Proprietary
2
1. Usos de la
estadística para
entender el mundo
digital
Google Confidential and Proprietary
4
Estadísticas globales y para España
Google Confidential and Proprietary
5
Uso de dispositivos
1. How many Mobile
users are there vs other
devices users?
2. Who is the Mobile
user? When, where and
why is it Mobile used?
Consumer Connected Survey TNS
Our Mobile Planet Ipsos
3. How is the Mobile
used within a Multiscreen
context?
4. What is the Mobile role
in the Purchase Journey
Multiscreen Survey Ipsos
Mobile Purchase Journey Ipsos
Google Confidential and Proprietary
Global consumers move to multi-screening
Number of connected devices per household
average: 1.5
Q: Which, if any, of the following devices do you currently use?
(Smartphone, PC, laptop Webbook, Tablet, Internet TV)
Source: TNS, Global Enumeration Study, 2012
Base: All respondents 16+; 1,000 per country
Google Confidential and Proprietary
7
PC / Laptop usage is high in most countries
Q1 2012 PC or laptop usage
average: 70%
Q: Which, if any, of the following devices do you currently use?
Source: TNS, Global Enumeration Study, 2012
Base: All respondents 16+; 1,000 per country
Google Confidential and Proprietary
8
Strong growth for Smartphones worldwide
2011 - 2012 Smartphone usage
X1.33
Q: Which, if any, of the following devices do you currently use?
Source: Global Enumeration Study, Ipsos 2011, TNS 2012
Base: All respondents 18+; 1,000 per country
Google Confidential and Proprietary
9
Strong growth for Tablets worldwide
2011 - 2012 Tablet usage
X1.6
Q: Which, if any, of the following devices do you currently use?
Source: Global Enumeration Study, Ipsos 2011, TNS 2012
Base: All respondents 18+; 1,000 per country
Google Confidential and Proprietary
10
Hihger Intent to use tablets (except in Africa)
In Africa, smartphones are expected to grow faster than tablets
Question: Which, if any, of the following devices do you currently use?
Question: You are currently not using a so called [device]. How likely are you to buy or use a [device] within the next 12 months?
Source: Global TNS Infratest enumeration study Q1 2012
Base: All respondents 16+; 1,000 per country
Google Confidential and Proprietary
11
Africa: mobile connections more important
than fixed connections
Index on total population
Question: What type of internet connection is your main internet connection at home, i.e. how is your home connected to the internet ?
Source: Source: Global TNS Infratest enumeration study Q1 2012
Base: All internet users with home connections 16+; Africa 8639 Asia 4,897; Europe 16,456; NA 1,673; SA 716; Oceania
1,728
Google Confidential and Proprietary
12
Tablet usage appears to be complementary
to other devices
Source: Spain Consumer Connected Survey Q1 2013
Base: All respondents 16+; 1,000
Google Confidential and Proprietary
PCs for productive activities, mobile for intimate usage.
Tablets still to be defined (enterteinment now)
Leisurely
Curious
Friendly
Versatile
Intense
Sexy
Casual
Impulsive
Quick
Fast
Efficient
Accessible
Productive
Dependable
Powerful
Important
Smart
Spontaneous
Social
Intimate
Essential
Google Confidential and Proprietary
Preferred Device is related to the activity
Working
Enterteinment
Commuting
Lunch
% Relative Uses by 24 hours /day
Source: Multiscreen Survey Ipsos 2013
Google Confidential and Proprietary
Uso de smartphone
Source: Our Mobile Planet Ipsos 2013
Google Confidential and Proprietary
Interacción entre las distintas pantallas
Not at the same time
85%
1. Sequential Usage
Moving from one device to another at different times
to accomplish a task (ie. booking a summer travel)
At the same time
2. Multitasking Usage
Using more than one device at the same time for an unrelated
activity (ie. Watching a TV show and playing Angry Birds with a
smartphone app)
3. Complementary Usage
Using more than one device at the same time for a related activity
(ie following a soccer match on TV and getting other matches
updates on a website in the tablet)
-
78%
22%
Universe of reference: 19.7 Mill of Smartphone, PC & TV users aged 16 or more
Source: Multiscreen Survey Ipsos 2013
Google Confidential and Proprietary
89%
of smartphone users have
looked for local
19,248,624
information
77%
ind
14,821,441 ind
have taken action as
a result
32%
6,159,560 ind
have visited the business
27%
5,197,129 ind
have called the business or
service
How often do you look for information about local businesses or services on your smartphone? (Ever)
Base: Private smartphone users who use the Internet in general and who look at least less than once a month
for information on their smartphone, Smartphone n= 891 Which of the following actions have you taken after
having looked up this type of information (business or services close to your location)?
Source: Our Mobile Planet Ipsos 2013
Google Confidential and Proprietary
Mobile research to Desktop purchase
Mobile users
Researched on smartphone
and purchased on desktop
41%
Mobile, PC & TV users
Start shopping on smartphone
and continue or finish on desktop
91%
6.7 M started an online purchase via smartphone, 6.1 M converted via PC;
5 M started planning a trip via smartphone, 4.5 M finally converted via PC
Source: Our Mobile Planet Ipsos 2013
Source: Multiscreen Survey Ipsos 2013
For the 41%: Listed below are various products or services. For each of these products or services please indicate which statement applies to you. “Research on smartphone then purchased via
computer” Source: OurMobilePlanet
Fort the 91%: Base: Have Started Activity on One Device & Continued on Another (488). Q. For the activities listed below, think about the last time you started each activity on one device and then
Google Confidential and Proprietary
continued or finished the same activity on another device. Please select which device you started and then continued on. If you have not done this, select “I have not done this”.
Search dominates the sequential usage
Search again on the second device
94%
Directly navigating to the destination site
88%
Via email / sending a link to myself
82%
Universe of reference: 19.7 Mill of Smartphone, PC & TV users aged 16 or more
Source: Multiscreen Survey Ipsos 2013
Base: Have Started Activity on One Device & truly Continued on Another: Searching (249); Browsing (296); Shopping (224), Watching a Video (181). Q. You mentioned
that you have started each activity below on one device and then continued it on another device. For each activity (column), please indicate the way(s) in which you did
Google Confidential and Proprietary
this.
Conclusiones
1
La representatividad, consistencia y continuidad es clave
en el análisis estadístico de tendencias sociales.
2
La investigación de mercados basada en datos
declarativos sirve para entender el pulso social, las
tendencias de los mercados, y comparativas entre
países (pues son metodologías al alcance de casi todos)
3
Para cuantificar la realidad del uso de dispositivos,
consumo de medios y audiencias de medios deben
desarrollarse mediciones más sofisticadas basadas en
“datos comportamentales” (y no “declarativos”) en la
medida que las nuevas tecnologías lo ponen cada vez
más al alcance del investigador.
Google Confidential and Proprietary
23
Usos de la estadística
para mejorar el
marketing
Google Confidential and Proprietary
24
¿Cómo hace dinero Google?
Companies pay Google for advertisements that
appear on the Google search results page
56%
60%
59%
53%
51%
Companies pay Google for the best possible
ranking in the search results
44%
45%
41%
30%
44%
Companies pay for information that Google
collects about you
35%
28%
28%
26%
19%
Companies pay Google to place their
advertisements on other pages across the web
34%
39%
42%
42%
30%
Google makes money in a different way
11%
10%
7%
10%
6%
I don’t know how Google makes money
29%
28%
30%
35%
27%
Ipsos MediaCT January to February 2012
Source: Q24a. In your opinion, which, if any, of the following best describes how Google makes money?
Base: All respondents n=2,991
Google Confidential and Proprietary
25
Google Confidential and Proprietary
26
Push
Google Confidential and Proprietary
27
Pull
How much you’re willing to pay (bids)
How “good” your ads are (relevance)
Competition
Google Confidential and Proprietary
28
La estadística aplicada a la comprensión de las
necesidades del consumidor: Search
Google Confidential and Proprietary
29
La estadística aplicada a la comprensión de las
necesidades del consumidor: Google Analytics
30
La estadística aplicada a la comprensión de las
necesidades del consumidor: True View
Preroll vs. TrueView at equal efficiency
2.8x
Source: GfK (randomised experiment with 18 brands and n = 3.184 respondents)
Connected Life Panel Germany 2012 (n = 10.000 panelists)
Google Confidential and Proprietary
31
La estadística aplicada a la comprensión de las
necesidades del consumidor: Retargeting
Google Confidential and Proprietary
32
Herramientas públicas: Google Trends
Google Confidential and Proprietary
33
Herramientas públicas: Google Trends
DIETA
Google Confidential and Proprietary
34
Herramientas públicas: Google Trends
DUKAN
Google Confidential and Proprietary
35
Herramientas públicas: Google Trends
-DUKAN
Google Confidential and Proprietary
36
Herramientas públicas: Flu Trends
Google Confidential and Proprietary
Herramientas públicas: Bank of England
Google Confidential and Proprietary
Conclusiones
1
La libre elección de información ha cambiado el modo en
el que el consumidor está dispuesto a acceder al
contenido publicitario.
2
La estadística tiene un papel clave para conectar las
necesidades del consumidor al contenido y formato
publicitario que está dispuesto a aceptar y valorar en
última instancia.
3
Existen herramientas públicas accesibles para todos que
permiten entender y optimizar nuestra oferta, y ejemplos
que nos inspiran al uso de la estadística para mejorar la
gestión de múltiples productos y servicios.
Google Confidential and Proprietary
39
Uso de la estadística
para medir la
efectividad del
marketing
Google Confidential and Proprietary
40
La escalabilidad es clave
Building
Scalable
Tools
Improving
research
standards
Improving
marketing
products
Rolling
out
externally
Google Confidential and Proprietary
El consenso también
Google Confidential and Proprietary
42
Tres dimensiones fundamentales
To demonstrate & empower
digital effectiveness
within Media Mix
Reach
Branding
Sales
Google Confidential and Proprietary
Reach: via fusión de datos fija
KANTAR MEDIA TV PANEL (n: 11.912)
Official Spain TV currency
KANTAR WORLPANEL households (n: 8.400)
KANTAR WORLDPANEL individuals (n: 4.863)
Exposure to Online based
on impression tags
FIXED
FUSION
Exposure to TV based on
fusion to TV currency panel
(Kantar Media)
•
Fusion process is based on similarities among Kantar Media (TV) and Kantar Worldpanel (online) panelists.
•
“Twins” are created based on sociodemographics, attitudes, TV chanel profile, day band profile (multidimensional reduction) and 200 additional
exclusive clusters.
44
Google Confidential and Proprietary
Reach: via fusión de datos fija
Only YouTube added an average of 2.9 extra reach points to TV
Average TV
reach (1+)
Average YT reach
(1+)
79.1%
14.3%
Only TV
67.7%
Overlap TV
+ YouTube
11.4%
Exclusive
Youtube reach:
+2.9 p%
Source: Kantar Worldpanel cross media analysis– internet users Base: 10 YouTube campaigns in 2012/2013
45
Google Confidential and Proprietary
Reach: via fusión de datos fija
More than 20% of total YouTube reach is not doubling up with TV . YouTube share of incremental reach goes
higher while looking at effective TV reach
Media Reach by different levels of TV OTS (%)
YouTube share of
incremental reach
(100= Total YouTube
reach)
20.2%
41.2%
55.9%
Source: Kantar Worldpanel cross media analysis– internet users Base: 10 YouTube campaigns in 2012/2013
46
Google Confidential and Proprietary
Reach: via fusión de datos fija
WEB media is particularly effective at delivering additional GRP´s among lower TV exposed audience
Almost 75% of online GRPs delivered out of high TV exposed
Source: Kantar Worldpanel cross media analysis– internet users Base: 12 campaigns in 2012/2013
47
Google Confidential and Proprietary
Reach: via Single Source Panel
• What we did:
 Analyzed 80 TV campaigns for different products of leading advertisers
 TV media plans by Ebiquity, YouTube media plans simulated by Google
• Data source:
 Single-source panel: Measurement of individual TV & online behavior from
same consumers in 5,000 households within GfK Media Efficiency Panel
Google Confidential and Proprietary
48
Reach: via Single Source Panel
For illustration only
reallocation of wasted TV OTS to unexposed
audiences
Google Confidential and Proprietary
49
Reach: via Single Source Panel
Benefits of Share-Shift from a Budget-Point of View
Share Shift
TV + YOUTUBE
TV ONLY
Stable Reach
NET REACH: 69%
NET REACH: 69%
COSTS
COSTS
Source: Share shift analysis Google, based on GfK MEP and Ebiquity data
Google Confidential and Proprietary
50
Brand Impact: ejemplo preroll vs skippable
Looking at recall, awareness and recognition no major differences in impact can
be seen between the two formats. If any, prerolls perform sligthly stronger.
% of respondents per cell
indicates differences at 90% significance level to control
Source: GfK / nurago 2011, n = 1.023 interviewss. Audi and Redbull Ads
Nurago Impact Panel Germany (n = 10.000 panelists)
Google Confidential and Proprietary
51
Brand Impact: randomized experiments
IMPACT PANEL (10.000 panelists, used for ad effectiveness research). Unique
methodology: nurago technology allows experimental study designs for ad effectiveness
in a real live campaign setting
-
Full control over reach and frequency of campaign delivery
Experiment embedded into the panelists‘ natural online usage
Campaigns can be fully simulated or boosted if needed within the panel
Survey
Panelists
(representative for
onliners 18-69 years)
Random pre campaign
assignment into test and
control groups, e.g.
Panelists fall into groups
depending on their
natural online behaviour
# only allowed to see skippable
# only allowed to see preroll
# did not see advertising
# exposed to skippable
# exposed to preroll
Google Confidential and Proprietary
Brand Impact: eficiencia preroll vs skippable
However, in terms of efficiency (impact per GRP) skippable prerolls clearly
outperform standard prerolls by factor 2.4 on average across brand metrics
Uplift to control per GRP
Efficiency
Index
234%
Efficiency
Index
248%
Efficiency
Index
234%
Source: GfK / nurago 2011, n = 1.023 interviewss. Audi and Redbull Ads
Nurago Impact Panel Germany (n = 10.000 panelists)
Google Confidential and Proprietary
53
Ventas: Econométricos “online to store”
Google Confidential and Proprietary
08/01/06
05/02/06
05/03/06
02/04/06
30/04/06
28/05/06
25/06/06
23/07/06
20/08/06
17/09/06
15/10/06
12/11/06
10/12/06
07/01/07
04/02/07
04/03/07
01/04/07
29/04/07
27/05/07
24/06/07
22/07/07
19/08/07
16/09/07
14/10/07
11/11/07
09/12/07
06/01/08
03/02/08
02/03/08
30/03/08
27/04/08
25/05/08
22/06/08
20/07/08
17/08/08
14/09/08
12/10/08
09/11/08
07/12/08
04/01/09
01/02/09
01/03/09
29/03/09
26/04/09
24/05/09
21/06/09
Vtas_Brutas_Adsl_Total_Resi_TLMK
21/06/09
17/05/09
12/04/09
08/03/09
01/02/09
28/12/08
23/11/08
19/10/08
14/09/08
10/08/08
06/07/08
01/06/08
27/04/08
23/03/08
17/02/08
13/01/08
09/12/07
04/11/07
30/09/07
26/08/07
22/07/07
17/06/07
13/05/07
08/04/07
04/03/07
28/01/07
24/12/06
19/11/06
15/10/06
10/09/06
06/08/06
02/07/06
28/05/06
23/04/06
19/03/06
12/02/06
08/01/06
Ventas: Econométricos “online to store”
1. Need to collect all the
variables that may influence
the sales
2. An analyst uses
judgement and SAS to
produce different regresion
models until best
combination is identified
80.000
70.000
60.000
50.000
40.000
30.000
20.000
10.000
-10.000
0
-20.000
-30.000
3. So he is able to identify
each variable influence to
the sales
80.000
70.000
60.000
50.000
40.000
30.000
20.000
10.000
0
4. Final check: the blue line
is the reality (sales); the red
line is the model (how it fits
the sales)
Ajuste
Google Confidential and Proprietary
Ventas: Econométricos “online to store”
8.9%
18.3%
2.4%
1.1%
1.1%
1.8%
1.5%
1.4%
Web
visits
8,9%
0.5%
12% of visit to dealer are
influenced by media.
26.3%
TV 1,5%
Outdoor
1,4%
36.6%
Magazines 0.5%
Number of
Present Unemployment 2000 E Plan
working days Situation Index
rate
VAT increase
Seasonality
Exogenous
TV advertising
Outdoor
advertising
Magazines
advertising
Web traffic
Dsplay
Organic Search
Paid Search
Web visits
Visit to dealer
Repeated ealer visit
Purchase
Other Media
Google Confidential and Proprietary
Ventas: Econométricos “online to store”
Paid search is the most
cost-effective.
100
ROI: Visits to dealer
by 1€ invested
Online ROI
Offline ROI
43
35
25
TV Fiat
500
1
3
TV Fiat
Punto
TV Fiat
Bravo
4
21
TV Fiat Magazines Outdoor Display
Gama
Fiat 500
F500
Fiat 500
Monetary value of an application and actual returns not shown due to client confidentiality – ROI data indexed
Paid
Search
Fiat 500
Display
Organic Search
Paid Search
Web visits
Visit to dealer
Repeated ealer visit
Purchase
Other Media
Google Confidential and Proprietary
57
Ventas: Econométricos “online to store”
*Historical Media Mix
*Improved Modelled Media Mix
Paid SearchCine Otros
0.3% 0.7% 4.0%
Display
7.8%
TV
43.1%
Exterior
18.7%
Radio
0.4%
Revistas
10.8%
Prensa
14.1%
Google Confidential and Proprietary
Conclusiones
1
La integración de la medición de audiencias y coberturas
desde múltiples dispositivos y plataformas es el principal
reto actual en la medición de efectividad publicitaria.
2
El desarrollo de las nuevas tecnologías en la emisión de
contenido publicitario permite la optimización y
escalabilidad de la medición del impacto publicitario en
notoriedad e imagen de marca.
3
Es fundamental explorar la medición del ROI de las
diferentes actividades de marketing de un modo
integrado y comparable entre ellas. La creación de
modelos econométricos es una vía sólida aunque
costosa en tiempo, dinero e infraestructuras de gestión
de la información interna.
Google Confidential and Proprietary
59
¿Preguntas?
Google Confidential and Proprietary
60
Descargar

Modelos Econométricos para La ONCE