putting science behind the standards - ipg media lab · © 2015 ipg media lab. proprietary &...
TRANSCRIPT
© 2015 IPG Media Lab. Proprietary & Confidential
HOW BIG IS THE VIEWABILITY PROBLEM?
2Source: Integral Ad ScienceStatistic for the U.S.
57%Of display ads are not
human/viewable
© 2015 IPG Media Lab. Proprietary & Confidential
WE HAVE A STANDARD TO GO BY…
3
STANDARD BANNER ADS RICH MEDIA ADS VIDEO ADS
At least 50% in viewFor a minimum of 1 second
At least 30% in viewFor a minimum of 1 second
At least 50% in viewFor a minimum of 2 consecutive seconds
WE STILL DON’T KNOW HOW EFFECTIVE ADS THAT MEET THE MEDIA RATINGS COUNCIL’S (MRC) VIEWABILITY STANDARDS ACTUALLY ARE.
4
© 2015 IPG Media Lab. Proprietary & Confidential
1.
What value do ads that meet the MRC minimum
viewability standard actually offer?
2.
What is the relationship between viewability and
ad effectiveness?
3.
How can marketers get the biggest bang
for their buck?
TO FILL IN SOME OF THESE BLANKS, WE EXPLORED…
5
© 2015 IPG Media Lab. Proprietary & Confidential
OUR GOAL IS TO…
6
Present results from the first known study that scientifically ties viewability to ad effectiveness by testing the following hypotheses:
OUR GOAL IS NOT TO…
Recommend changes to the MRC standard, or develop a new viewability standard
1. The more viewable an ad is, the more consumers will see it2. As viewability increases, so does ad effectiveness3. There are strategies advertisers can employ to make less viewable ads more effective
© 2015 IPG Media Lab. Proprietary & Confidential
RIGOROUS SCIENTIFIC TESTING
7
Participants recruited from nationally representative
online panel
Randomized into 1 of 189 viewability test cells; Viewed
webpage that matched their typical consumption habits
Subset of sample viewed webpage while being eye-tracked
Answered post-exposure survey for branding metrics
© 2015 IPG Media Lab. Proprietary & Confidential
189 DIFFERENT AD SCENARIOS WERE TESTED
8*Two standard display ad sizes tested; 1 per industry vertical**Logo placement test cells only included for 1 ad type for 1 brand
PERCENT IN VIEW TIME IN VIEW AD TYPE INDUSTRY VERTICAL LOGO PLACEMENT AUDIO SHARE OF VIEWCONTEXTUAL RELEVANCE
25% .5 secStandard
Banner AdCPG Top On 1 of 1 ads In Context
30% 1 secRich Media/
Large Format Ad
Auto Not at top Off 1 of 2 adsOut of
Context
50% 2 sec Video Ad 1 of 4 ads
75% 4 sec
100% 7 sec
Full Exposure
© 2015 IPG Media Lab. Proprietary & Confidential
CONTROLLING FOR VIEWABILITY
9
“ PERCENT IN VIEW ”
Ad was never more/less than the designated % in view
“ TIME IN VIEW ”
Test ad rotated to a house ad after thedesignated time frame
© 2015 IPG Media Lab. Proprietary & Confidential
1.
Viewability is highly related to ad effectiveness
2.
How long the ad is in view matters more
than how much of it is in view
3.
All ad strategies are not created equal: some help more at lower viewability levels than others
KEY FINDINGS
10
© 2015 IPG Media Lab. Proprietary & Confidential
UNDER MRC STANDARD AT MRC STANDARD ABOVE MRC STANDARD
NATURALLY, HIGHER VIEWABILITY = MORE EYEBALLS
12
Consumers see ad
23%Consumers
see ad
48%
Consumers see ad
76%
Results include all ad types based on eye tracking data
© 2015 IPG Media Lab. Proprietary & Confidential
EYEBALLS DON’T NECESSARILY = AD RECALL
13Results include all ad types for comparable test cells
While attention increases with viewability, that doesn’t guarantee consumers are internalizing the ad
25%
51%
74%
17% 17%
32%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Under standard Meets MRC standard Above Standard
CONSUMER ATTENTION VS. AD RECALL
% Who Looked At Ad % Ad Recall
© 2015 IPG Media Lab. Proprietary & Confidential
MRC STANDARD ISN’T A MAGICAL THRESHOLD FOR AD EFFECTIVENESS
14
* Statistically significant difference between control and test at >=90% confidence^ Statistically significant difference between MRC standard and test at >=90% confidenceResults include all ad types
Ads that simply met the standard did not have impact on ad recall
+2%
+16%*^
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
AD RECALL
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
(D
ELT
A)
AD RECALL BY MRC STANDARD (DELTA)
Met MRC Standard Above Standard
© 2015 IPG Media Lab. Proprietary & Confidential
WHICH MEANS EVEN SOME IMPRESSIONS BELOW THE MRC STANDARD HAVE IMPACT
15
* Statistically significant difference between control and test at >=90% confidence^ Statistically significant difference between MRC standard and test at >=90% confidenceResults include all types
+6%*^
+2%
+16%*^
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
AD RECALL
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
(D
ELT
A)
AD RECALL BY MRC STANDARD (DELTA)
Under Standard Met MRC Standard Above Standard
© 2015 IPG Media Lab. Proprietary & Confidential
6%*^
Under standard
BELOW MRC STANDARD BREAKOUT BY EFFECT ON AD RECALL
MOSTLY BECAUSE SOME PARTIAL ADS COULD BE IN VIEW FOR A LONG PERIOD OF TIME
% VIEW-belowTIME-meets MRC
Results include all ad types* statistically significant difference between control and test at >=90% confidence^ statistically significant difference between % Met MRC standard and % Under standard at >=90% confidence
% VIEW-meetsTIME-below MRC
% VIEW-aboveTIME-below MRC
% VIEW-belowTIME-BelowMRC
% VIEW-belowTIME-above MRC
16
© 2015 IPG Media Lab. Proprietary & Confidential
16%*^
Above standard
ABOVE MRC STANDARD BREAKOUT BY EFFECT ON AD RECALL
ON THE FLIP SIDE, SOME ADS ABOVETHE STANDARD DON’T HAVE IMPACT
Results include all ad types* statistically significant difference between control and test at >=90% confidence^ statistically significant difference between % Met MRC standard and % Under standard at >=90% confidence
% VIEW-aboveTIME-above MRC
% VIEW-atTIME-above MRC
Specifically, when time in view is low
% VIEW-aboveTIME-At MRC
% VIEW-aboveTIME-At MRC
17
© 2015 IPG Media Lab. Proprietary & Confidential
BUT, WHAT IS MOST IMPORTANT IS THAT VIEWABILITY IS HIGHLY RELATED TO EFFECTIVENESS
18
* Statistically significant difference between control and test at >=90% confidenceResults include all ad types, % in view under the MRC standard excludes large format since it was not tested
5%4%
9%*
11%*
0%
10%
20%
Under standard At MRC standard 75% 100%
PER
CEN
T C
HA
NG
E IN
REC
ALL
(D
ELT
A)
PERCENT IN VIEW
AD RECALL BY PERCENT IN VIEW (DELTA)
Averaged Delta for All Ad Types
1%
3%
8%*
17%*
0%
10%
20%
Under standard At MRC standard 4 seconds 7 seconds
PER
CEN
T C
HA
NG
E I
N R
ECA
LL (
DELT
A)
TIME IN VIEW
AD RECALL BY TIME IN VIEW (DELTA)
Averaged Delta for All Ad Types
© 2015 IPG Media Lab. Proprietary & Confidential
“TIME IN VIEW” IS KING
19
*This was only tested for Standard Banner and Video Ads, since Large Format was not tested with a lower % in viewResults include all ad types
TIME IN VIEW
PER
CEN
T IN
VIE
W
7 sec0
100%
EFFECT OF PERCENT AND TIME IN VIEW ON AD RECALL (DELTAS)
Percent in view higher, Time in view lower
+3.8%Percent in view higher, Time in view higher
+14.0%
Percent in view lower, Time in view lower
+0.4%Percent in view lower, Time in view higher*
+10.4%
AT MRC STANDARD
© 2015 IPG Media Lab. Proprietary & Confidential
NEED TO GIVE PEOPLE TIME TO SEE THE AD
20Results based on eye tracking data – includes all ad types
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
At MRC standard Above standard
PER
CEN
T SEEN
PERCENT IN VIEW
PEOPLE WHO SAW AD BY PERCENT IN VIEW (DELTA)
Averaged Delta for All Ad Types
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Under standard At MRC standard Above standard
PER
CEN
T S
EEN
TIME IN VIEW
PEOPLE WHO SAW AD BY TIME IN VIEW
Averaged Delta for All Ad Types
© 2015 IPG Media Lab. Proprietary & Confidential
-5%
0%
5%
10%
15%
20%
25%
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
PER
CEN
T C
HA
NG
E I
N A
D R
ECA
LL (
DELT
A)
AD RECALL BY TIME IN VIEW (DELTA)
Banner Ads Rich Media Ads Video Ads
GENERALLY, YES.
“TIME IN VIEW” IMPACTS ALL AD TYPES SIMILARLY
TIME IN VIEW (Seconds)
22
© 2015 IPG Media Lab. Proprietary & Confidential
BECAUSE LARGE FORMAT ADS HAVE THE LOWEST VIEWABILITY STANDARDS, THEY ALSO HAVE THE MOST TO GAIN FROM INCREASED VIEWABILITY
23
13% 15%
50%
28%
54%
77%
59%
84%
96%
0%
20%
40%
60%
80%
100%
120%
BANNER ADS LARGE FORMAT VIDEO
PER
CEN
T O
F U
SER
S
PERCENT SEEN BASED ON EYE-TRACKING DATA
Under standard Met MRC standard Above standard
NUMBERS BASED ON HOW MANY LOOKED AT THE AD AT SOME POINT DURING THE WEBSITE VISIT
Greatest growth
© 2015 IPG Media Lab. Proprietary & Confidential
0%
5%
10%
15%
20%
25%
25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100%
PER
CEN
T C
HA
NG
E
IN A
D R
ECA
LL (
DELT
A)
AD RECALL BY PERCENT IN VIEW (DELTA)
Banner Ads Video Ads
AUDIO-ON VIDEO EFFECTIVENESS DOES NOT CHANGE AFTER 75% IN VIEW
24*Video includes audio on
Although banner ads are measurably more effective after 75% in view, video ads do not have any measurable change
PERCENT IN VIEW
© 2015 IPG Media Lab. Proprietary & ConfidentialResults based on regression modelingN = 4,451
PREDICTED CHANCE CONSUMERS WILL RECALL STANDARD BANNER AD
• 19% chance of recall at the standard
• % chance over doubles from lowest to highest viewability
TIME IN VIEW
PER
CEN
T IN
VIE
W
31%
30%
29%
32%
25%
24%
25%
21%
21%
21%
20%
16%
19%
0.5 Sec 1 Sec 4 Sec 7 Sec
100%
75%
50%
25%
38%
19%
21%
26
© 2015 IPG Media Lab. Proprietary & Confidential
25%
Results based on regression modelingN = 1,953
PREDICTED CHANCE CONSUMERS WILL RECALL LARGE FORMAT AD
• 17% chance of recall at the MRC standard
• % chance over doubles from lowest to highest viewability
TIME IN VIEW
PER
CEN
T IN
VIE
W
35%
13%
32%
21%
15%
19%
21%
11%
22%
14%
18%
0.5 Sec 1 Sec 4 Sec 7 Sec
100%
75%
50%
30%
39%
11%
17%
18%
27
© 2015 IPG Media Lab. Proprietary & ConfidentialResults based on regression modelingN = 2,762
PREDICTED CHANCE CONSUMERS WILL RECALL VIDEO AD
• 10% chance of recall at the MRC standard
• % chance over triples from lowest to highest viewability
TIME IN VIEW
PER
CEN
T IN
VIE
W
34%
26%
32%
25%
24%
10%
12%
13%
17%
8%
13%
12%
9%
1 Sec 2 Sec 4 Sec 7 Sec
100%
75%
50%
25%
30%
34%10%
28
© 2015 IPG Media Lab. Proprietary & Confidential
NO SURPRISE, AUDIO MAKES VIDEO MORE EFFECTIVE
30
AB = Statistically significant difference between A/B at >= 90% confidence * = Statistically significant difference between test and control >= 90% confidenceN=1,261, Results for video
+8%*
+22%*^A
0%
5%
10%
15%
20%
25%
Ad Recall
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
VIDEO AUDIO ON VS OFF EFFECT ON AD RECALL (DELTA)
Audio off (A) Audio on (B)
275% lift for
video with audio on
© 2015 IPG Media Lab. Proprietary & Confidential
OVER 2X THE IMPACT FOR ADS UNDER THE MRC THRESHOLD
31
AB = Statistically significant difference between A/B at >= 90% confidence * = Statistically significant difference between test and control >= 90% confidenceN=1,261, Results for video
+5%+4%
+20%*
+14%*^A
+6%
+35%*^A
0%
5%
10%
15%
20%
25%
30%
35%
40%
UNDER STANDARD AT MRC STANDARD ABOVE STANDARD
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
VIDEO AUDIO ON VS. OFF BY VIEWABILITY LEVEL (DELTA)
Audio off (A) Audio on (B)
175% lift for ads under the
standard alone
• Audio on is especially helpful when there are fewer pixels in view
© 2015 IPG Media Lab. Proprietary & Confidential
IN GENERAL, PLACING LOGO AT TOP IS RECOMMENDED
32
* = Statistically significant difference between control and test at >= 90% confidenceN= 1,743Results for standard banner ads
+4%
+10%*
0%
2%
4%
6%
8%
10%
12%
AD RECALL
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
(D
ELT
A)
EFFECT OF LOGO PLACEMENT ON AD RECALL (DELTA)
Not at Top Top
Provides a63% lift in
effectiveness
© 2015 IPG Media Lab. Proprietary & Confidential
LOGO PLACEMENT STARTS HELPING IMMEDIATELY AT THE STANDARD
33Results for standard banner ad
-2%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
PER
CEN
T C
HA
NG
E I
N A
D R
ECA
LL (
DELT
A)
AD RECALL BY TIME IN VIEW (DELTA)
Not At Top Top
-2%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
25% 50% 75% 100%
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
(D
ELT
A)
AD RECALL BY PERCENT IN VIEW (DELTA)
Not At Top Top
TIME IN VIEW (Seconds) PERCENT IN VIEW
© 2015 IPG Media Lab. Proprietary & Confidential
A LESS CLUTTERED PAGE HELPS INCREASE BOTH AD AND MESSAGE RECALL
35
* = Statistically significant difference between control and test at >= 90% confidenceN= 2,603Results for standard banner ads
+6%
+3%
+6%
+2%
+10%*
+5%*
0%
2%
4%
6%
8%
10%
12%
AD RECALL MESSAGE RECALL
PER
CEN
T IN
AD
REC
ALL
(D
ELT
A)
EFFECT OF SHARE OF VIEW ON AD RECALL (DELTA)
1 of 4 Ads 1 of 2 Ads 1 of 1 Ads
© 2015 IPG Media Lab. Proprietary & Confidential
GREATER SHARE OF VIEW HAS IMPACT AFTER 1 SECOND IN VIEW
36Results for standard banner ads
-5%
0%
5%
10%
15%
20%
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
PER
CEN
T C
HA
NG
E IN
AD
REC
ALL
(D
ELT
A)
AD RECALL BY TIME IN VIEW (DELTA)
1 of 4 ads 1 of 2 ads 1 of 1 ads
TIME IN VIEW (Seconds)
WHILE SOME IMPRESSIONS UNDER THE STANDARD HAVE IMPACT, WE MUST CONSIDER:
38
MRC reports that there is a 77.2% chance that
impressions that meet the standard will ultimately
become 100% in view (all pixels in view).
© 2015 IPG Media Lab. Proprietary & Confidential
PAY DIFFERENT AMOUNTS FOR DIFFERENT LEVELS OF VIEWABILITY
40
Under standardno value
$0 Under standardwith value
$5At standard
$10
Above standard
$15
© 2015 IPG Media Lab. Proprietary & Confidential 42
ADS THAT DO EXCEED THE STANDARD END UP HAVING THE GREATEST IMPACT, AND MOST ADS THAT MEET THE STANDARD ALSO EXCEED IT.
VIEWABILITY IS IMPORTANT, BUT IS NOT THE END-ALL-BE-ALL. IT ISN’T A KPI.
SO…
© 2015 IPG Media Lab. Proprietary & Confidential
Use video ads, prioritizing ad spaces where audio is
likely to be on
In general, it doesn’t pay to only strive for
100% in view. Focus on ad placement where
ads are more likely to be viewed for longer periods of time (e.g. Out-Stream, Email)
Be sure to place the logo at the top of the
ad, where it is immediately visible
Aim for ad spaces that are less likely to be
cluttered with other ads
TO GIVE ADS THE BEST FIGHTING CHANCE WHEN VIEWABILITY MAY BE LOWER:
43
© 2015 IPG Media Lab. Proprietary & Confidential
NEXT STEPS
44
Cost analysis through
in-market testing
Test the effectiveness of
in-feed auto-play video
Explore the role of creative ad
quality