xcmo 2013: online & offline analytics convergence (peanut butter & chocolate)

23
#xcmo13 Peanut Butter & Chocolate Online & Offline Analytics Convergence October 10, 2013

Upload: cross-channel-marketing-optimization-group

Post on 01-Nov-2014

1.175 views

Category:

Technology


0 download

DESCRIPTION

For most brands, digital marketing lives in its own world, yet we know there are a multitude of offline marketing and external factors than impact results. Adometry's Tim McDonough provides a road-map and examples of bringing media mix modeling together with digital marketing attribution to better understand real results.

TRANSCRIPT

Page 1: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Peanut Butter & ChocolateOnline & Offline Analytics Convergence

October 10, 2013

Page 2: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Marketing Research

Page 3: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Marketing Research

› For more than a century marketing research had been done mainly with surveys

– 1879 - N. W. Ayer conducts custom research in an attempt to win the advertising business of Nichols-Shepard Co., a manufacturer of agricultural machinery.

– 1911 can be considered the year marketing research becomes an industry. That year, J. George Frederick leaves his position as editor of Printer’s Ink to begin his research company, the Business Bourse.

– Also in 1911, Kellogg's ad manager, R. O. Eastman creates the Association of National Advertisers which is now known as the Association of National Advertising Managers. The group’s first project is a postcard questionnaire to determine magazine readership.

How we got here

Page 4: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Marketing Research

– 1915, J. Walter Thompson establishes a Research Department– 1922, Dr. Daniel Starch tests reader recognition levels of magazine and

newspaper advertisements and editorial content. – 1923, Dr. George Gallup begins measuring advertising readership.– Also in 1923, Arthur Charles Nielsen founds the ACNielsen company to

conduct marketing research.› 1942, Nielsen inaugurates a National Radio Index for broadcasters and

advertisers.› 1950, Nielsen begins their television ratings service.

– The 1960s usher in the era of qualitative focus groups and sociologist Daniel Yankelovich assembles MONITOR, a large longitudinal panel that has been tracked continuously since its inception.

How we got here

Page 5: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Marketing Research

– In the 1970s computers collect and store historically massive amounts of data and the discipline of database analytics emerges.

– The 1980s bring behavioral psychology to the arena of brand research and cable TV revolutionizes broadcast media.

– In the 1990s trends in in-market tracking include a greater focus on the multimedia nature of entire advertising campaigns. CRM emerges.

– The 21st century opens with a decade of global reach with the dawn of the Internet age. The dichotomy of digital and traditional advertising emerges.

– Now in the second decade of the century attention returns to the complete marketing ecology. Online and offline brand engagement blurs into a continuum.

– 2007, JWT establishes its Marketing Science practice in Dallas. Recruiters from the traditional full service agencies are all now seeking “marketing scientists”.

How we got here

Page 6: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

State of the Advertising Industry

Page 7: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

US Advertising Media Market Sizes

Source: PricewaterhouseCoopers

TV Internet Newspaper Radio Magazine B2B OOH Video gamesTheaters0

10

20

30

40

50

60

70

80

90

63.8

36.6

22.8

16.1 16.4 14.1

7.5

0.8 0.7

81.6

69.4

18.4 17.3 15.211.9 9.6

1.5 0.9

2012 2017

$Bill

ion

Page 8: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

US Advertising Media Market Share

Source: PricewaterhouseCoopers

TV Internet Newspaper Radio Magazine B2B OOH Video games

Theaters0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

40.0%35.7%

20.5%

12.8%

9.0% 9.2%7.9%

4.2%

0.4% 0.4%

36.1%

30.7%

8.1% 7.7% 6.7%5.3% 4.3%

0.7% 0.4%

2012 2017

Page 9: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Agency Revenues

WPP (London)

Omnicom (NY)

Publicis (Paris)

Interpublic (NY)

Dentsu (Tokyo)

Havas (Paris)

Hakuhodo (Tokyo)

0

2

4

6

8

10

12

14

16

18

15.6

14.3

7.76.7

3.1

1.81.2

$Bill

ion

Source: Rediff

Page 10: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Marketing/Advertising Agencies

› Full service agencies› Creative agencies› Specialized agencies

– Demographic segments– Industry specific– Single client agencies

› Digital agencies– Web and social property design– SEO– Social

› Research agencies– Primary research (surveys & focus groups)– Database– Online analytics– Marketing analytics

› Media buyers

It’s a jungle out there

JWT

Page 11: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Digital Divide

Page 12: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Digital Divide

› The Internet has enabled us to track individual consumer behavior from first ad exposure to conversion and has created opportunities to engage consumers with brands interactively as never before.

› This has spawned a new category of research agencies devoted to analyzing the start to end online journey.

› Until recently, this behavior was described and modeled as something independent of exposure to all other marketing treatments.

› Clients have pushed back with business questions about how the conversion behavior is influenced by the marketing ecology as a whole.

Meet in the middle

Page 13: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

CMO Wish List

Page 14: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Digital Divide

› Agencies have responded to this pressure.

› Traditional marketing agencies are in a mad scramble to acquire or organically grow digital expertise.

› Digital agencies are doing the same with traditional marketing.

1 + 1 = 3

Page 15: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

The Digital Agency Challenge

N

ii

i

N

ii

xy

yyS

1

221

2

121

)(

)ˆ(),(

Page 16: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

The Traditional Agency Challenge

To increase his chances of getting a job with a big brand agency with his degree in Advertising, Arnold enrolls in Machine Learning 101.

Page 17: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Playing Together

Page 18: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Holistic Model Solution

Page 19: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Channel Level Overlay

Page 20: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Symbiosis

› The key to success in connecting top-down aggregate data models that operate at the channel level and user-level data models is to use information from each to augment the other.

› Top to bottom– Top-down aggregate econometric models provide the view of the ecology at the channel

level.– This information can be used to scale the results of bottom-up models to the rest of the

channels to provide absolute attribution to the user-level aggregates.

› Bottom to top– Bottom-up machine learning processes can provide noise reduction to the econometric

models in the form of refined variable specifications.– This noise reduction enhances the discrimination power of all channel variables in the

model.

Two-way information flow

Page 21: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Econometric Models are Here to Stay

› Since the advent of the computer in the late 1950s econometrics has transformed the business landscape.

› Many techniques dating back to the dawn of the applied economics age are still successfully employed in today’s markets.

– Single equation models– Simultaneous equation models– Time-series models

› Lag models› Vector autoregressive (VAR) models› ARIMA with transfer functions

› The machine learning community needs to become comfortable with their new colleagues who come to them from the social sciences.

Long history of success

Page 22: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Artificial Intelligence

› Machine learning is a mature discipline.

› Supervised and unsupervised data mining and modeling techniques have proved their worth.

› Social scientists need to learn how to become comfortable with their new computer science colleagues.

Short but solid history of success

Page 23: XCMO 2013: Online & Offline Analytics Convergence (Peanut Butter & Chocolate)

#xcmo13

Who can say no?