digital media and analytics - merkle inc....digital media and analytics how to leverage a powerful...
Post on 31-Mar-2020
4 Views
Preview:
TRANSCRIPT
@MerkleCRM #digex2012 Join the Conversation
Digital Media and Analytics
How to leverage a powerful new set of
techniques for digital media buying and program
measurement to drive business results
@MerkleCRM #digex2012 Join the Conversation
CRM Drives Competitive Advantage
2
We believe market leaders drive
competitive advantage through the ability to
understand, influence, measure consumer
behavior at the individual level but at mass
scale.
@MerkleCRM #digex2012 Join the Conversation
Risk and Opportunity…
• Growth Digital media spend is big and growing proportionately bigger
• Capability The technology has dramatically leapfrogged the current model
• Skillsets The data driven know how to exploit that advancement already exists
3
@MerkleCRM #digex2012 Join the Conversation
The Market Shift
4
Cu
rren
t Market
Toolsets
Knowhow
Toolsets
Knowhow
Me
dia A
d Te
ch&
CR
M K
no
wh
ow
Competitive Advantage
@MerkleCRM #digex2012 Join the Conversation
Merkle Media Trading Desk
Merkle CDI
Capabilities
DYNAMIC CREATIVE
DSP
ATTRIBUTION
DMP
AD VERIFICATION
Merkle Media Trading Desk
INVENTORY
Advanced Retargeting Data Validation Services
Lead Scoring
DATA PROVIDERS
Exchanges/SSPs
Data Providers
Direct Publishers
5
@MerkleCRM #digex2012 Join the Conversation
Media Planning
6
Display Media Plan
One
Platform
Cross sell & upsell
Prospecting
Winbacks
Display Media Plan
Many
Vendors
Ad.com
Collective Media
Tribal Fusion
Old Media Planning New Media Planning
@MerkleCRM #digex2012 Join the Conversation @MerkleCRM #digex2012 @MerkleCRM #digex2012 Join the Conversation 7
The Engine
Ari Buchalter, Caltech Astrophysicist PhD
@MerkleCRM #digex2012 Join the Conversation 8
Serving a Single Ad
• 10s-100s of variables
• Highly nonlinear
• Dynamic
• Real-time
Geo
User History
Frequency
Offline History
User data
Publisher Time of Day
Channel
Ad Size
Media data
Day of Week
Creative
Message
Offer
Ad data
Predicted Response Rate x Buyer’s Goal Value = Bid CPM
Over 20B impression opportunities per day in
display alone!
@MerkleCRM #digex2012 Join the Conversation
The Science behind the Engine
9
An algorithmic machine-learning approach is necessary to optimize display media
@MerkleCRM #digex2012 Join the Conversation 10
Revving the Engine within a Channel
Connecting the dots
across ads is as important as knowing what each
impression is worth
The fact that users see multiple ads opens up other optimization opportunities:
• Exposure tracking
• Message sequencing
• Dynamic creative
• Partial attribution
@MerkleCRM #digex2012 Join the Conversation 11
Full Throttle: Multi-Channel Optimization
Beyond 2012, even more channels are addressable
(DOOH, IPTV, etc.)
Display Search
Mobile Social
As we move beyond 2011, optimization within one channel won’t be enough, as users interact with brands across digital channels
Additional focus is needed on:
• Cross-channel tracking
• Multi-channel optimization
• Cross-channel attribution & decisioning
@MerkleCRM #digex2012 Join the Conversation @MerkleCRM #digex2012 @MerkleCRM #digex2012 Join the Conversation 12
The Analytics
Peter Vandre, MIT Economist
@MerkleCRM #digex2012 Join the Conversation
Effective Measurement
13
New Media Targeting= Rocket Science
Now some advice to actually start measuring this stuff • Get started- basic improvements on last click attribution • Cross media testing • Attribution modeling • Tackle offline to online
@MerkleCRM #digex2012 Join the Conversation
Better Direct Attribution
14
View Thru 1
2nd Click
1st Click
View Thru 2
60%
20%
15%
5%
1. Play around with your attribution assumptions—How sensitive are they?
2. Focus on refining view thru window assumptions
3. Analyze and test interactions across media types
Analysis Priorities
Last Click
Look Familiar? Closer to Reality
Your Other Media
@MerkleCRM #digex2012 Join the Conversation
Testing to Improve Measurement
Goal: Estimate Incremental impact of paid search terms
Situation: • Major retailer • High % of spend on
branded paid search • How much is really
incremental?
Outcome:
• Estimate 40% cannibalization between Paid and organic search
• Adjust keyword values accordingly in targeting
Shut off SEM Compare to post period
Incremental= Est. organic orders- actual organic
15
Solution • Branded keyword
blackout test
@MerkleCRM #digex2012 Join the Conversation
Modeling to Improve Measurement
16
Goal: Estimate Incremental performance for each digital campaign
Situation: • Major Financial
Services Company • Huge brand spend • What is ROI of digital
media?
Outcome: • Ongoing incremental
campaign estimates • Rebalancing between
top/bottom of funnel
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Incremental Impact
Search
Agg Lead Gen
Agg Ins
Video
Display
Solution: • Attribution modeling • Campaign level reporting
@MerkleCRM #digex2012 Join the Conversation
Modeling Offline and Online Interactions
17
Goal: Estimate Incremental performance for each digital campaign
Situation: • Financial Services
Company • Large DM program • Wanted to see if
Display could lift DM
Outcome: • Able to measure display lift on
DM program • 17% lift on DM with display
Solution: • Controlled test • Offline to online direct
match • Targeted display and DM
Display Responsive Audiences
DM Responsive Audiences
•95% of US Population •Auction Marketplace
RTB Inventory
Pull Prospect Lists
Database
Target in media DM Only
DM + Display
Display Only
Analyze Results 1 2 3
@MerkleCRM #digex2012 Join the Conversation @MerkleCRM #digex2012 @MerkleCRM #digex2012 Join the Conversation 18
Impact
Megan Pagliuca, Ad Tech Evangelist
@MerkleCRM #digex2012 Join the Conversation
Increased Efficiency
19
0%
10%
20%
30%
40%
50%
60%
70%
80%
0%
10%
20%
30%
40%
50%
60%
70%
1 6 11 16 21 25+
Users / Impressions by Frequency for July
Unique Users by Frequency
Undercontacting
•66% of the users see only
one impression
Metric Total Percent
Average # Impression 5
Total Unique Users 10,000,000
Total Impressions/ Month 50,000,000 100%
Optimal Impressions 25,000,000 50%
Questionable Impressions 15,000,000 30%
Waste Impressions 10,000,000 20%
Consolidate Buys to One Platform
Overcontacting
•20% of impressions were being shown
to users with at 50+ frequency
Example 1: Example 2:
@MerkleCRM #digex2012 Join the Conversation
Decreased CPA
20
201
221213
200218
169
107 113101
104 119
$-
$10,000
$20,000
$30,000
$40,000
$50,000
$60,000
$70,000
$80,000
$90,000
$-
$50.00
$100.00
$150.00
$200.00
$250.00
$300.00
Targeted Display CPA vs. SpendSpend CPA
Integrated Credit Card Approval Data into Algorithm
Situation: • Fi Serve Advertiser • Decrease cost per acquisition
of new credit card approvals
Outcome: • 5x improvement when offline
data was incorporated in algorithm
Integrated credit card
approval
(Indexed)
@MerkleCRM #digex2012 Join the Conversation
Increased Volume High Value Accounts
21
Built Integrate Multi-Channel Acquisition (I.M.A.) Program
CPA
Traditional Agency Program Merkle Data Driven Program
Situation: • Fi Serve Advertiser • Increase # of high value
checking accounts • Increase online and branch
account opens
TIME
Outcome: • 217% decrease in Cost per
New Account • 143% increase in percent of
High Value/Low Value Responders at same budget level
@MerkleCRM #digex2012 Join the Conversation
7 Things You Can Do To Improve Your Digital
Media Program Tomorrow
1. Work with the media trading desk that best suits your needs
2. Centralizing execution into a single targeting platform
3. Do not work with partners that don’t give you transparency
4. Ensure site and email data collection allow you to collect offline data at
every opportunity
5. Pass offline conversion and/or value data into online algorithm
6. Ensure measurement & analysis is being done at the impression level,
using just a dsp or ad server aggregate reporting does not give you the
full picture
7. Start testing and thinking about your media budget based on
incremental impact
22
@MerkleCRM #digex2012 Join the Conversation
Questions???
23
@MerkleCRM #digex2012 Join the Conversation
Round Table Questions
– Where does the responsibility for Media sit in your organization? How is it connected to CRM activities?
– Are you using an Agency Trading Desk or is your agency working directly with publishers
– Have you started consolidating your data-driven buys? If so, why not?
– What is something you can take back to your office this week to take advantage of some of these changes in the landscape?
– Are you using display to drive offline actions? Why or why not?
– Are you integrating back end metrics into display targeting?
24
@MerkleCRM #digex2012 Join the Conversation
Ari Buchalter
COO, Mediamath 646.840.4203
abuchalter@mediamath.com
Megan Pagliuca
VP, Digital Media .212.790.1569
mpagliuca@merkleinc.com
Peter Vandre
VP, Digital Analytics 443.542.4345
pvandre@merkleinc.com
top related