cross channel attribution: path to maximize media...
TRANSCRIPT
©2016, Analytic Mix, Inc. All data herein is sourced to Analytic Mix, Inc., unless otherwise noted.
Cross Channel Attribution:Path to Maximize Media ROI
Haren Ghosh
President & CEO
Analytic Mix Inc.
www.analyticmix.com
November 2016
©2016, Analytic Mix, Inc.
Analytic Mix (AM) provides advertising measurement and optimization solutions to
advertisers and media agencies
HQ in Silicon Valley, R& D center in Milwaukee, technology center in India, and sales
offices in Los Angeles, Atlanta, and New York
We specialize in
Enterprise-level marketing resource management
Automated Analytics and Dashboard Data Delivery
Digital and Cross Channel Media Mix Modeling
Campaign Performance Measurement
Predictive Analytics and Forecasting
Predictive Channel Attribution Modeling
Client value –
Clients use our analytic insights to make decisions about how to allocate their marketing dollars in terms of spend, media mix, creative execution, and audience targets to ensure the highest possible ROI.
About Analytic Mix (AM)
2Strictly Confidential
©2016, Analytic Mix, Inc.
Analytic Mix (AM) Enterprise Level Offerings
Campaign Level Cross Media AnalysisWe integrate survey responses and answer the following,
What is the optimal media mix for each campaign?
What’s the most efficient media to reach target audiences?
What is the point of diminishing return?
Cross-Channel Attribution Modeling
Our approach accounts for
Both immediate and carry-over effects of media channels Non-linear effects of media channels on brand KPIs Optimal allocation across dayparts, stations, etc. Synergistic effects of other media channels on brand KPIs Paid, owned, and earned media
Strategic Marketing Mix Modeling (MMM)
We include disparate data assets such as
Brand and category sales Competitive landscape and analytics Planned and actual media spend Macroeconomic variables Pricing and promotional data
3Strictly Confidential
15% 18%
54%
88%
Macro Drivers Marketing Budget Structural Factors Overall
Strength of Marketing Mix Model
Equivalent to 264M Dollars
©2016, Analytic Mix, Inc.
Analytic Mix Media ROI Measurement Platform
TV
Spots
Stations
Dayparts
Local/National
Cable/Network
Title
Genre
Magazine/
Newspaper
Online
Display
Video
Search
Website
Analytic Mix Big Data
Management and
BI Platform
Advertising Programs
Media Data
Cost
GRPs/TRPs
Circulations
Impressions
Competitive
Spend
Sales
Orders/$$
Registrations
Site Visitations
Lead Gen systems
Ad Effectiveness on Sales Optimal Spend Optimal Media Mix
Clie
nt
Dat
a
Age
ncy
Dat
a
Other
Game Console
Events
Mobile
Out of Home
Rep
ort
ing
4Strictly Confidential
©2016, Analytic Mix, Inc.
Analytic Goals & Research Parameters
Analytic Goals
o Determine total media budget needed to maximize media attributable brand KPI
o Establish maximum weekly spend limits to help guide weekly media buy
o Recommend variations in media spend based on seasonal patterns and promotional events
Research Parameters
o Data Range: Daily data from January, 2015 to May, 2016
o The following data assets were utilized for the current analysis:
• Media spend data
• TV, Digital, and Radio
• Promotional calendars
• Email drop dates
• Brand KPI data
• Competitive media spend
6Strictly Confidential
©2016, Analytic Mix, Inc.
Key Finding 1: Media Altogether Drive ~65% of the Brand KPI
7Strictly Confidential
o The brand’s effective media channels jointly drive about 65% of the KPI.
o While all media channels in place work in synergy, TV has the highest contribution to the brand.
• The effect of TV is higher among new customers.
©2016, Analytic Mix, Inc.
Key Finding 2: Reallocation of TV Investment Across Dayparts Generates Higher Sales
o A gradual adjustment of TV investment across dayparts following the model generated recommendations enables the brand to achieve higher sales even when the overall TV budget is unchanged.
8Strictly Confidential
Dayparts Details Historical Allocation Recommended Allocation
Early Morning M-F, 6am-9am 14.6% 8.7%
Daytime M-F, 9am-4pm 6.9% 9.7%
Early Fringe M-F, 4pm-7pm 12.1% 7.0%
Prime Time M-N, 7pm-11pm 31.7% 33.2%
Late Fringe M-N, 11pm-2am 7.7% 10.9%
Overnight M-N, 2am-6am 0.9% 4.0%
Weekend Sa-Sun, 6am-7pm 26.0% 26.4%
©2016, Analytic Mix, Inc.
Key Finding 3: Media Effects Follow a Diminishing Pattern and Maximize at $330K Weekly Media Budget
o As media budget increases, the volume of media attributable brand KPIs increases. At $330K media budget, the number of media attributable brand KPIs reaches the maximum point. Beyond the maximum point, the effect of current media strategy will potentially generate a negative effect caused by over exposures.
9Strictly Confidential
-1%
4%
9%
14%
19%
24%
1,200
1,300
1,400
1,500
1,600
1,700
1,800
1,900
2,000
2,100
2,200
2,300
2,400
2,500
Med
ia A
ttri
bu
tab
le K
PI G
row
th R
ate
Med
ia A
ttri
bu
tab
le B
ran
d K
PI/
Wee
k
Weekly Media Budget
Revenue Response Function Attributable to Media
KPI Growth Opportunity KPI Change Rate
Maximum Point –Weekly Max: ~ $330KMedia Attributable KPIs: 2,476Growth Rate: 0%
Spend in Apr’16 – ~$265KMedia Attributable Kits: 2,182KPI Growth Rate: 5.5%
©2016, Analytic Mix, Inc.
Key Finding 4: Execution of Modeling Recommendations Helps Achieve ~12% Additional Sales
o As the current media activities followed model generated recommendations closely, the sales trajectory has shifted upward by ~12%. The current model with updated information predicts an increase in orders as media budget reaches to an weekly average of $330K. At the maximum point, incremental order attributable to $10K weekly TV budget becomes nearly zero.
10Strictly Confidential
1,800
1,900
2,000
2,100
2,200
2,300
2,400
2,500
$260K $270K $280K $290K $300K $310K $320K $330K $340K
Med
ia A
ttri
bu
tab
le B
ran
d K
PI/
Wee
k
Weekly Media Budget
Revenue Response Function Attributable to Media
Apr'16 Mar'16
Sales Trajectory Prior to Model Implementation
Sales Growth Following Model Generated Recommendations
©2016, Analytic Mix, Inc.
Key Finding 5: Media Exposure Produces Carry-Over Effects, Which Diminish to an Insignificant Level After 28 Days
o The effectiveness of media on number of registrations diminishes with time. The data suggest that the media effects last for 28 days for the current high involvement product and reduce by a steady rate every day it passes.
• As media exposure produces delayed effects on registrations, it’s unlikely to establish any relationship between registrations and hourly media activities. However, when media activities are aggregated at a sufficiently higher level (i.e., a week), the relationship between those variables becomes prominent.
11Strictly Confidential
23
0 4
37
62
4 79
2 94
3 1,0
79
1,2
01
1,3
11
1,4
10
1,5
00
1,5
80
1,6
52
1,7
17
1,7
76
1,8
28
1,8
76
1,9
19
1,9
57
1,9
91
2,0
23
2,0
51
2,0
76
2,0
98
2,1
19
2,1
37
2,1
54
2,1
69
2,1
82
-
500
1,000
1,500
2,000
2,500
Cu
mu
lati
ve B
ran
d K
PI A
ttri
bu
tab
le t
o M
edia
Media Effectiveness Span
Carry-Over Effects of Advertisement on Orders
Cumulative KPI
Lag = 28 DaysDecrement = 9%
©2016, Analytic Mix, Inc.
Key Finding 6: TV Allocation Across TV Stations Vary by Dayparts
o To come up with a robust optimal allocation strategy for more than 150 TV stations and several dayparts, we have developed a systematic modeling approach and included all data inputs at a daily level. Our model produces optimal (i.e., recommended) allocation scores for all TV stations per dayparts and week. Further, the model guides what should be the weekly budget and how that budget should be allocated across TV stations and dayparts.
• The table below shows only a small subset of all TV stations.
Dayparts TV StationsCurrent Allocation,
%Recommended Allocation, %
CPR -AnalyticMIx
Current Allocation, $
Recommended Allocation, $
Day Time American Heroes 0.020 0.005 6.38 $10,000 $2,371
Day Time Food Network 0.007 0.003 9.57 $3,500 $1,580
Day Time H2 0.003 0.021 1.44 $1,500 $10,466
Prime Hallmark Channel 0.019 0.000 $9,500 $0
Prime Headline News 0.004 0.015 1.98 $2,000 $7,638
Prime HGTV 0.004 0.000 $2,000 $0
Weekend History 0.019 0.002 17.82 $9,500 $849
Weekend Lifetime 0.019 0.002 17.82 $9,500 $849
Weekend LMN 0.019 0.003 8.91 $9,500 $1,697
12Strictly Confidential
© 2016, Analytic Mix, Inc.
Thank you!
Haren GhoshPresident & CEO
Analytic Mix Inc.www.analyticmix.com
1-888-675-6290