Please see the full video of the study presentation here: https://vimeo.com/75477600
- Using facial coding to measure creative excellence
- What emotional patterns have been winning the hearts and minds of the jury in the last three years?
- Can we predict Cannes Lions winners going forward?
- How can we use this every day to make better decisions in content creation and media buying?
Mihkel Jäätma - Managing Director, Realeyes
Sander Saar - Product Manager, AOL
3. Source: Gunn Report 2000-2008, IPA
Les Binet & Peter Field,
Marketing in the Era of
Accountability, 2007
Emotional advertising
campaigns are more
effective and more
profitable than rational
campaigns - even in
'rational' categories…
“
”
11x
more efficient
in market share
growth
The Gunn Report
14. Strongest Variables
Don’t drop
the game
Use
sadness
early
Make
women
happy at
the end
Disgust
and
recover
Make
women
excited at
the end Don’t start
with
confusion
Time the
peaks right
Grow the
trend
15. Make women happy at the end
Female
Male
0:05 0:10 0:15 0:20 0:25 0:30 0:35
Peak in
last third
(32%) of
the video
SHORTLIST
16. Don’t drop the game
Happy
0:05 0:10 0:15 0:20 0:25
NOTHING
Avoid low
happiness after
first third of
the video
22. Time the peaks right
GOLDMEDALNOTHING
SHORTLIST
End 0.8
0.7
0.6
0.5
0.4
0.3
Start 0.2
Max Happy
Min Happy
23. Don’t drop
the game
Use
sadness
early
Make
women
happy at
the end
Disgust
and
recover
Make
women
excited at
the end Don’t start
with
confusion
Time the
peaks right
Grow the
trend
Modelling Cannes
We looked at the
8 strongest rules.
Now let’s expand
that set to 500
24. Importance of different emotions
Positives are
more important
in predicting
shortlists
Negative emotions
are necessary to
predict creative
excellence GOLDMEDAL
SHORTLIST
100%
80%
60%
40%
20%
0
Sad
Scared
Confused
Surprised
Happy
Neutral
25. 0.8
0.7
0.6
0.5
0.75
0.66
0.63
Can we predict Cannes in future?
YES
shortlist with
75% accuracy!
In the future,
medals will need
more data to get
a solid prediction
model
AUC
GOLDMEDAL
SHORTLIST
26. Strongest Variables
Don’t drop
the game
Use
sadness
early
Make
women
happy at
the end
Disgust
and
recover
Make
women
excited at
the end Don’t start
with
confusion
Time the
peaks right
Grow the
trend
27. Strongest Variables
Don’t drop
the game
Use
sadness
early
Make
women
happy at
the end
Disgust
and
recover
Make
women
excited at
the end Don’t start
with
confusion
Time the
peaks right
Grow the
trend
28. Attraction
Hook in 8 seconds
Retention
Keep the audience
Impact
Kahneman’s Peak-End
Engagement
Connect via emotions
EmotionAll®
29.
30. Confusion is
dominant
emotion for first
45 secs, lowering
attraction
Happiness
spikes and
dominates
throughout
Happiness
drops, but video
ends with
peak, leaving
viewers with
strong impact
Minimal
negative
emotions
throughout
31. This video has very good
overall performance.
Better than 95% of videos.
So Real it’s Scary
Emotional
Impact
33. Conclusions
Confusion in the first
45 seconds lowers
attraction. Happiness
spikes and dominates
throughout.
Top
5%
Compared with
video A, LG leaves
viewers with a
higher impact.
A B
• 18-34
• Single, divorced,
separated.
• Low or very low
income.
~
34. Get attention with a strong spike of
surprise and happiness in the first eight
seconds. Avoid confusion.
Prevent drops and end
with a spike for the
strongest impact.
Ending with an engagement
spike benefits the syndication
strategy of publishers with
higher play-to-end rates.
Syndicate to relevant
audiences with the
highest engagement.
Learning
€
Activate the video
across UK and US
audiences
35. 15” to 5’
content length
20+
categories
EmotionAll®
1,000+
videos tested
30+
countries
36. Consumers love emotional content...
3xhigher average
view to end %
8xhigher click
through rate
20xbetter in
converting views
into social
actions
100xQuicker in
attracting
viewers
Source: Be On research
Inefficient in ROIWHY All working on media sideShould we look at creative?Do not know the impact of creative WHY
In addition to IPA, also Cannes
Entered > Shortlisted > Medals > Golds each year visually and talk about all the different juries at those stages briefly?Lots of numbers and methodology for cannes LionsCarefully constructed creative selectionThere is no way to measure creative effectively WHY
DEMO
Add bubbles around the graph:Avg, Max, Min% ofCohortsSmoothing settingsTrendsEventsHead poseGesturesBy AgeBy Gender
Picked from distinctions between three key groupsShortlisted vs Not ShortlistedMedal vs RestGold vs RestThe point is to show which groups were compared to identify strongest drivers of performance.
Build them in one by one on click in the following order:Make Women Happy In The EndDon't Drop The GameDisgust And RecoverUse Sadness EarlyDon’t Start With ConfusionMake Women Excited In The EndGrow The TrendTime The Peaks RightArrange all the bubbles around the golden
Maybe even without the sceen, just lines in full, you call.
Future improvementsMore dataMore meta-variables Match audience with videos Segment by genre Session level variables
Build them in one by one on click in the following order:Make Women Happy In The EndDon't Drop The GameDisgust And RecoverUse Sadness EarlyDon’t Start With ConfusionMake Women Excited In The EndGrow The TrendTime The Peaks RightArrange all the bubbles around the golden
Build them in one by one on click in the following order:Make Women Happy In The EndDon't Drop The GameDisgust And RecoverUse Sadness EarlyDon’t Start With ConfusionMake Women Excited In The EndGrow The TrendTime The Peaks RightArrange all the bubbles around the golden
5 countries 3 ads in 1 hour – no time for such depth, need simpler
Data from LG report
Emotionally stronger ads get3x higher average view to end %8x higher click through rate20x better in converting views into social actions100x quicker in attracting viewers.Source: Be On and Realeyes research based on 400 videos study, 2013
… imagine if 15 seconds into watching the ad we can already tell what the brand lift and purchase intent of the viewing will be without asking a single question