More Related Content Similar to Are you leaving ROI on the table? Similar to Are you leaving ROI on the table? (20) More from Yan Rozovsky (8) Are you leaving ROI on the table?1. Are You Leaving ROI on the Table?
Lessons Learned
Best Practices for Maximizing Campaign Results in Challenging Times
Evan Neufeld, Vice President, Advertising Effectiveness
May 2009
2. Agenda
– About Us …
The Current Environment
Implications for Measuring Digital Advertising Effectiveness
Benchmarking Success via Comprehensive Norms
© comScore, Inc. Proprietary and Confidential. 2
3. comScore Customer Knowledge Platform:
A 360°
View of 2 Million Global Internet Users
WEB VISITING & VIEWING DEMOGRAPHICS
• All Web Site/Page Click Stream • Self-reported and validated
• Content Viewed • Appended Segments (e.g. Claritas, Acxiom)
• Search Engine Queries • Individual & Household Level
• Keyword Used
ONLINE TRANSACTIONS SURVEYS
• All Secure Session Activity • E-mail or Contextual “Pops”
• Purchases and Subscriptions • Behavior-activated Surveys
• Price Paid, Shipping & Handling, • Observation of All Surveys Taken
Promotions Across All Suppliers
• Applications/Configurations
MARKETING STIMULI OFFLINE PURCHASING
• Linked using Name and Address
• Online Ads TV VIEWING • Client CRM Databases
• Referral Links • Link to Digital Set Top TV Data
using name and address • Retailer Loyalty Card Data
• IRI Scanner Panel Data
Designed to be representative of the online population
Projectable to the total US population
TRUSTe certified for information privacy & security
© comScore, Inc. Proprietary and Confidential. 3
4. How It Works: Using a Test and Control Design
The randomized experimental design creates campaign exposed
(Test ad cell) and non-exposed groups(Control ad cell)
Both groups are invited and surveyed with
the same screeners and incentives via our
proprietary AdRecruit technology
Exposed to Ads
(Test Cell)
Not exposed to Campaign Ads
(Control)
© comScore, Inc. Proprietary and Confidential. 4
5. Benefits of Real-World Test/Control + Panel: Observance of the
Campaign Live In-Market
comScore Panelists
A sub-set of the Test and
Control will naturally fall to
the comScore panel
This allows us to provide:
Behavioral metrics such as
Exposed to Ads
(Test Cell) lift in search and visitation
TotalLift™ Calibration
Inclusion of select Post – Buy
Metrics
Accurate demographics thru
elimination on sample bias
Not exposed to
Campaign Ads
(Control)
© comScore, Inc. Proprietary and Confidential. 5
6. Brand Metrix ™ : Proof That Online Ads Work
Includes branding effects!
• Brand awareness and attitudes
• Brand purchase intent
• Advertising awareness and message recall
Includes behavioral metrics!
• % Lift in Site Visitation
• % Lift in Brand Search
Includes key campaign metrics!
• Reach and frequency,
• Gross rating points (GRPs)
• Demographics
© comScore, Inc. Proprietary and Confidential. 6
7. comScore Brand Metrix ™ : Over 300 Ad Effectiveness Studies
Across Multiple Industries
Media Agencies Financial Retail Pharma CPG/Restaurant Travel Tele/Tech Auto
© comScore, Inc. Proprietary and Confidential. 7
8. Agenda
About Us …
– The Current Environment
Implications for Measuring Digital Advertising Effectiveness
Benchmarking Success via Comprehensive Norms
© comScore, Inc. Proprietary and Confidential. 8
9. The Internet is Important to at Least 6 in 10 Consumers, with 50%
Citing an Increase Compared to a Year Ago
Importance of Internet
Source: comScore Survey, January 2009
Q: How important has the Internet become Q: How has this changed versus
in providing you with information to help a year ago?
you make buying decisions?
65%
50%
Very Important
Significantly Increased
Slightly Important
Somewhat Increased
Past 3 months Versus a year ago
© comScore, Inc. Proprietary and Confidential. 9
10. Economic Hardship Leads to Dramatic Increase in
Financial/Economic-related Searches Across the Web
Number of Unique Searchers (MM) for Each Term: Dec. 2008 vs. YA
Source: comScore
Y/Y %
change +27% +58% +100% +83% +160% +157% +223%
© comScore, Inc. Proprietary and Confidential. 10
11. 2008 eCommerce Growth Substantially Lower Than Preceding Years
Online Consumer Dollar Sales Growth
Excluding Auctions, Autos and Managed Travel ($Billions)
Source: comScore
+7%
+17% $221
$200
+20%
+22% $171 +9%
+12%
+26% $143
+13%
+29% $117
+20%
$93
+26%
$72
+33%
+21% +6%
Travel +24%
+24%
+26% +26%
Non-
Travel
© comScore, Inc. Proprietary and Confidential. 11
12. All Income Segments Are Turning to Coupon Sites More so Than
They Did Last Year
Number of Unique Visitors to Coupon Sites and Growth vs. YA
Source: comScore
40,000 50%
+46% 45%
35,000
40%
30,000 +38%
Percent Growth
Percent Growth
Unique Visitors
35%
25,000
+33% +33% 30%
20,000 +29% +26% 25%
+21% 20%
15,000 +20%
15%
10,000
+11% 10%
5,000 +4%
5%
+4%
0 0%
Feb-2008 Mar-2008 Apr-2008 May-2008 Jun-2008 Jul-2008 Aug-2008 Sep-2008 Oct-2008 Nov-2008 Dec-2008
© comScore, Inc. Proprietary and Confidential. 12
13. Agenda
About Us …
The Current Environment
– Implications for Measuring Digital Advertising Effectiveness
Benchmarking Success via Comprehensive Norms
© comScore, Inc. Proprietary and Confidential. 13
14. Holistic Measurement More Important Than Ever…
Who? Premium Targeting
Who was reached and how often? Re-Target In-Market Consumers
• Reach/Frequency • Improve Campaign
• Impressions Effectiveness by hitting
the right audience
• Demographics
• Previous Shoppers
• Flighting
• Competitor Shoppers
• Audience Build
•Behavioral targets
• Distribution
•Etc.
Branding Impact What Happened?
Brand Metrix™ Lifts On & Offline Behavioral Lifts
•Site Visitation
•Brand Awareness •Searches
•Attitudes •Paid/Organic Clicks
•Message Recall •Offline Purchases
•Purchase Intent •“Retail Impact” &
•Offline Conversion “Close the Loop”
Studies
© comScore, Inc. Proprietary and Confidential. 14
15. The Majority of Advertising Often Falls Outside the Desired Target
Outside US
Frequency <=4
In US But Not Target
Hit Target with >=5 Frequency
100%
8%
90%
80%
70%
60% 61%
Area of improvement
50%
40%
30%
20% 19%
10%
12% Likely Waste
0%
Weighted Average - 8 Campaigns
© comScore, Inc. Proprietary and Confidential. 15
16. Think Beyond Demos Into Premium Targets:
7200 Segments Suitable for a Myriad Audience Targets
CPG 150 Categories, 500 + brands
Pharma RX and conditions
Auto Cars owned, date of purchase, New purchases
Online Shopping computers, cameras, TV’s, office supplies, furniture, apparel, …
Offline Shopping Retail, Restaurants, Grocery, Entertainment ….
Online and off-line Travel Air, hotel, car travel, cruises, ….
Personal Consumption 100 categories
Wireless Usage Mobile Apps and Devices
Credit card, insurance, mortgage applications, trading activities,
Financial
brokerage and banking information
Off-line Media Consumption TV, Magazine,…
Off-line Product Ownership Appliances, Electronics, …
© comScore, Inc. Proprietary and Confidential. 16
17. comScore study with Google:
“The Role of Search in Consumer Buying”
The Three Components of How Search Drives Buying
Direct Online Effects 16%
Latent Online Effects 21%
Latent Offline Effects 63%
Only measuring direct (i.e. immediate) online effects understates
ROI by a factor of 6X
Source: Google/comScore “The Role of Search in Consumer Buying” Press Release,
March 21, 2006; based on holiday-related purchases completed online and offline
© comScore, Inc. Proprietary and Confidential. 17 across 11 product categories for 60 days post-search
18. Objective Setting is Critical -- Which Needle(s) are You
Trying to Move?
Number of Individuals
Attitudinal
Increased Value of Individuals
Increased Value of Individuals
– Awareness
Awareness,
Perceptions &
Usage – Opinions
Brand
Differentiation & – Intent
Attribute
Association
Behavioral
– Site visitation
Target Future – Site engagement
Intent
– Share of mind
Preference
– Search
Purchase
– Online
Actual Purchase
– Offline
© comScore, Inc. Proprietary and Confidential. 18
19. The Cookie Deletion Problem: Overstates Campaign Reach but
Understates Frequency
30% of Internet users delete their cookies in a month
– This rate corroborated by independent studies from Belden, Jupiter, and others
These deleters do so an average of 4 times a month
– 5 different cookies for same site in a month on one computer
Cookie deletion creates major problems:
– 2.5 times overstatement of unique visitors in server logs
– 2.5 times overstatement of reach and a similar understatement of frequency in
ad server log
News: The deletion rate
has increased by about
25% in only two years
© comScore, Inc. Proprietary and Confidential. 19
20. Why is Cookie Deletion so Important with Regard to Measuring Ad
Effectiveness?
Problem: Since users delete cookies, current control groups
are contaminated and effectiveness is understated
comScore Panelists
Solution: TotalLift™ Calibration
– We use the comScore panel to determine control
group contamination
– Lift is adjusted upward, providing accurate
results
Exposed to Ads
(Test Cell)
Example 15% Lift
• Unaided recall is 15%
• comScore determines that Everyone Else
there was a 20% cookie
deletion rate
• comScore TotalLift™ 19.5% Lift
Not exposed to
algorithm reveals 19.5% comScore TotalLift™
Campaign Ads lift
(Control) © comScore, Inc. Proprietary and Confidential. 20
21. The Percent of Users Deleting Cookies Varies Across Ad Campaigns
Sample of cookie deletion percentages
30.0% across campaigns in different categories 26.6%
25.0% 23.7%
22.5%
19.8% 19.8%
20.0% 19.0%
17.6% 17.9%
15.0%
10.0% 8.6%
5.0% 3.9%
0.0%
Pharma
Pharma
Pharma
Non Profit
Tech
CPG
Finance
Finance
Retail
Retail
Under-estimation of branding effects can be significant
© comScore, Inc. Proprietary and Confidential. 21 Source: comScore Ad Effectiveness Studies
22. By Analyzing Cookie Deletion Data We Can Understand the True
Impact of the Campaign
Favorability towards BRAND X
50%
50% 45%
45% 45%
45% 45% 45%
40%
40% 38%
40% 38%
40%
% Top Box
% Top 22Box
30% 28%28%
30%
20% Control
Control
20%
10%
10%
∆ Lift +5 ∆ Lift +7 ∆ Lift +18 Test Test
% Lift: +12.5% % Lift: +18.4% % Lift: +60.1%
0%
0%
Favorability
Favorability Low Cookie Deletion HighHigh Cookie Deletion
Low Cookie Deletion Cookie Deletion
(3.9%)
(3.9%) (26.6%)
(26.6%)
Control: 327 Test: 313 Control: 327 Test: 313 Control: 327 Test: 313
Q. How would you describe your overall opinion of each of the following brands?
(scale of 1 to 7 where 1 = Very unlikely and 7 = Very likely)
© comScore, Inc. Proprietary and Confidential. 22 indicates a significant lift above control at 90% (CL)
23. Agenda
About Us …
The Current Environment
Implications for Measuring Digital Advertising Effectiveness
– Benchmarking Success via Comprehensive Norms
© comScore, Inc. Proprietary and Confidential. 23
24. Measuring the Holistic Impact of Online Ad Campaigns
300+ studies conducted to assess the impact of paid search and online ads
on online and offline sales
Real world analysis: comScore panelists divided into two matched groups
(exposed and non-exposed to advertising)
– Search only
– Display ads only
– Search and display ads together
– Neither
Passively measured behavior and / or surveys
– Linked to in-store buying through CRM databases, retailer loyalty cards (we
have 125 Million card dataset accessible), credit card data, IRI scanner panel
© comScore, Inc. Proprietary and Confidential. 24
25. Brand Metrix ™ Market Norms For Attitudes: Online Shows Positive
Impact Across Key Measures
Top-of-Mind 29.6 ∆ Lift +3.0
∆ Lift +3.0
Unaided % Lift: +10.1%
% Lift: +10.1%
Awareness 32.6
89.3 ∆ Lift +.8
Aided Awareness % Lift: +.9%
90.1
Total Advertising 29.7
∆ Lift +2.2
Awareness 31.9 % Lift: +7.4%
Favorability (Top 2 27.3 ∆ Lift +3.5
box) 30.8 % Lift: +12.8%
Control Test
© comScore, Inc. Proprietary and Confidential. 25
26. Brand Metrix ™ Market Norms Can Help Measure Impact Across
Properties...
The campaign drove directional lift in intent to recommend BRAND X, with
significant lift on Site B
Unaided Awareness of BRAND X
40% 38%
35% 32%
30%
30% 28% 28%
26%
% Top 2 Box
25%
20% Control
15% Test
10% ∆ Lift +4 ∆ Lift +12 ∆ Lift +2
5% % Lift: +14.3% % Lift: +46.1% % Lift: +7.1%
0%
Control: 412 Test: 393 Control: 91 Test: 80 Control: 321 Test: 313
Site A Site B Site C
Q. When thinking of types of Y what brand comes to mind first?
© comScore, Inc. Proprietary and Confidential. 26 indicates a significant lift above control at 90% (CL)
27. Brand Metrix ™ Market Norms Can Help Measure Impact Across Ad
Unit Types….
Both Video and Standard creative units achieved significant lifts showing
successful execution at both the basic and most advanced design levels
Favorability towards BRAND X
100%
85% 81%
80% 75%
66%
% Top 2 Box
61%
60% 56%
Control
40%
Test
∆ Lift +5 ∆ Lift +10 ∆ Lift +15
20% % Lift: +8.9% % Lift: +13.3% % Lift: +22.7%
0%
Rich Media Video Standard
Control: 151 Test: 189 Control: 117 Test: 293 Control: 184 Test: 769
Q. How would you describe your overall opinion of each of the following brands?
(scale of 1 to 7 where 1 = Very unlikely and 7 = Very likely)
© comScore, Inc. Proprietary and Confidential. 27 indicates a significant lift above control at 90% (CL)
28. Brand Metrix ™ Market Norms can help measure impact across
placements…
At the campaign level, all placements proved a success with the
Sponsorship achieving highest impact
Intent to Purchase BRAND X
60% 53%
50%
39%
% Top 2 Box
40% 36%
33%
30% 24% 26% Control
20% Test
∆ Lift +15 ∆ Lift +20 ∆ Lift +10
10%
% Lift: +62.5% % Lift: +60.1% % Lift: +38.5%
0%
ROS Sponsorship Roadblock
Control:181 Test: 487 Control: 206 Test: 233 Control: 321 Test: 313
Q. In the next 30 days how likely are you to purchase each of the following brands?
(scale of 1 to 7 where 1 = Very unlikely and 7 = Very likely)
© comScore, Inc. Proprietary and Confidential. 28 indicates a significant lift above control at 90% (CL)
29. Brand Metrix ™ Market Norms for Behavioral Data:
Driving Traffic to Advertiser Web Site
There is significant lift that would be overlooked by only considering immediate actions
The effects of online display advertising are persistent
Though additional benefits are gained over time the proportional value of the lift when compared to the
Control group (i.e. percent lift) lessen over time
Advertiser Site Reach
6.6%
5.8%
4.8%
4.5%
3.9%
3.5%
3.1%
2.1%
% Lift: 65.0% % Lift: 53.8% % Lift: 49.1% % Lift: 45.7%
Week of first exposure Weeks 1-2 after first exposure Weeks 1-3 after first exposure Weeks 1-4 after first exposure
Control Test
© comScore, Inc. Proprietary and Confidential. 29
30. Brand Metrix ™ Market Norms for Behavioral Data: Display Ads Lift
Site Visitation Across All Verticals
Advertiser Site Reach
Weeks 1-4 after first exposure
4.5% % Lift: 46%
Average, N=139 ∆ Lift: 2.1%
6.6%
0.9% % Lift: 114%
Automotive, N=38 ∆ Lift: 1.0%
1.9%
1.3% % Lift: 86%
Finance, N=16 ∆ Lift: 1.1%
2.3%
0.6% % Lift: 77%
CPG & Restaurant, N=10 ∆ Lift: 0.5%
1.1%
9.1% % Lift: 52%
Retail & Apparel, N=21 ∆ Lift: 4.7%
13.8%
7.0% % Lift: 42%
Media & Entertainment, N=24 ∆ Lift: 2.9%
10.0%
5.8% % Lift: 25%
Electronics & Sof tware, N=14
7.2% ∆ Lift: 1.5%
4.8% % Lift: 21%
Travel, N=9
5.8% ∆ Lift: 1.0%
Control Test
© comScore, Inc. Proprietary and Confidential. 30
31. Brand Metrix ™ Market Norms for Behavioral Data: Impact of Display
Ads on Search Still Evident 4 Weeks Post Initial Exposure
While a relatively small percentage of users exposed to an online display ad subsequently make a
search on an advertiser’s Trademark/Brand term the lift between test and control is significant
However, this dynamic is important because of the synergy between display ads and search (more on
this later) and because a Trademark or a Brand search is a significant indicator of purchase intent.
% Making a TM/Brand Search
0.9%
0.7%
0.6%
0.5%
0.5%
0.4%
0.3%
0.2%
% Lift: 52.3% % Lift: 46.0% % Lift: 40.3% % Lift: 38.1%
Week of first exposure Weeks 1-2 after first Weeks 1-3 after first Weeks 1-4 after first
exposure exposure exposure
Control Test
© comScore, Inc. Proprietary and Confidential. 31
32. Brand Metrix ™ Market Norms for Behavioral Data:
Clear Synergies Between Search & Display
Conventional wisdom says that media works best when multiple approaches are employed to work
together
Here we see clear synergies between search and display. In each case, both in terms of % reach and $$
spent per 000 exposed, it’s clear that the combination of search & display together outpace the sum of
the results of display and search separately
% Making A Purchase on the Advertiser Site
(Retail Only) 5.1%
2.4% 1.9%
1.0% 1.5% 1.1%
Display Only Search Only Search & Display
% Lift: +42% % Lift: +121% % Lift: +173%
Online $$ per 000 Exposed
(Retail Only)
$6,107
$2,724 $2,723
$994 $1,263 $1,548
Display Only Search Only Search & Display
% Lift: +27% % Lift: +76% % Lift: +124%
Control Test
© comScore, Inc. Proprietary and Confidential. 32
33. Brand Metrix ™ Market Norms for Behavioral Data:
Online is Important but What About Offline?
Higher offline sales lifts are found for search advertising vs. display but when combined the synergy
provides even more lift
Lift generated by online advertising is significant and would not be captured by a CPC or cookie-based
model
Incremental Impact on Offline Sales per (000) Exposed
Source: comScore
© comScore, Inc. Proprietary and Confidential. 33
34. Brand Metrix ™ Market Norms for Behavioral Data:
Reach of Display Ads Often Helps Lift Sales More Than Search
The overall number of people reached by Display ads is typically much higher than Search
As a result, even though the sales lift among those exposed to a Search ad is higher, the total dollar
sales gained from Display ads is often larger than Search due to the smaller lift acting on a larger base
Incremental Impact on Offline
% of Households Reached Sales per (000) Exposed Total Sales Lift Index
119% 200 198
81% 200% 198%
82%
100
100%
11% 8%
16%
Search & Search Only Display Search & Search Only Display Only
Display Only Search and Search Only Display Only Display
Display
© comScore, Inc. Proprietary and Confidential. 34
35. Don’t leave ROI on the table: 6 Key Takeaways
Focus on 360 degree campaign measurement
Look to target, enhance, and focus media buying
Set clear objectives & measure accordingly
Be sure to measure both the direct and latent impact of your
campaign on both attitudes & behavior
Exploit the synergies between media types (search, display, cross
media)
Be sure to correct / understand the impact of cookie deletion
© comScore, Inc. Proprietary and Confidential. 35