putting science behind the standards - ipg media lab · © 2015 ipg media lab. proprietary &...

© 2015 IPG Media Lab. Proprietary & Confidential

PUTTING SCIENCE BEHIND THE STANDARDS


HOW BIG IS THE VIEWABILITY PROBLEM?

2Source: Integral Ad ScienceStatistic for the U.S.

57%Of display ads are not

human/viewable


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


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


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


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


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


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


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

WHAT VALUE DO ADS THAT MEET THE MRC MINIMUM STANDARD

ACTUALLY OFFER


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


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


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


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


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


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


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)



“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


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)


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


DOES VIEWABILITY IMPACT ALL TYPES

OF ADVERTISING THE SAME WAY?

21


-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)


Banner Ads Rich Media Ads Video Ads

GENERALLY, YES.

“TIME IN VIEW” IMPACTS ALL AD TYPES SIMILARLY

TIME IN VIEW (Seconds)

22


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


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)


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

WHAT ARE THE CHANCES?

25

© 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


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

HOW CAN MARKETERS GET THE BIGGEST BANG FOR THEIR BUCK?

29


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


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


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


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)


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)


Not At Top Top

TIME IN VIEW (Seconds) PERCENT IN VIEW


WHAT ABOUT SHARE OF VIEW?

34


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


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)


1 of 4 ads 1 of 2 ads 1 of 1 ads

TIME IN VIEW (Seconds)

WHAT DOES THIS MEAN?

37

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).

IF WE REALLY WANTED TO DEVELOP STANDARDS TIED TO EFFECTIVENESS, WE WOULD…

39


PAY DIFFERENT AMOUNTS FOR DIFFERENT LEVELS OF VIEWABILITY

40

Under standardno value

$0 Under standardwith value

$5At standard

$10

Above standard

$15

BUT, THE MRC STANDARDS AREN’T MEANT TO GUARANTEE AD EFFECTIVENESS.

THAT’S THE AD’S

JOB.

41

© 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…


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


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


THANK YOU

putting science behind the standards - ipg media lab · © 2015 ipg media lab. proprietary &...

Documents