1) The document analyzes four viral videos or memes to understand what factors contribute to something going viral.
2) It finds that variables like views, shares, and engagement alone do not determine virality. Instead, content spreads either as "spikers" with fast, high peaks or "growers" with slower growth over a longer period.
3) The structure and fragmentation of the target audience impacts which spread pattern occurs, with more fragmented audiences correlating with slower, grower-type growth. Understanding audience communities is key to effective distribution strategies.
8. So given the right content, audience relevance and influencer
push, virality should always happen in the same way.
Except it never does
9. We looked at 4 memes that have “gone viral”:
a music video, an ad, a citizen journalism video, a web series
10. 0
10,000
20,000
30,000
40,000
50,000
60,000
11-May 18-May 25-May 01-Jun
Launched
at
10pm
GMT
on
12
May,
&
gets
11,400
Twi<er
shares
in
2
hours
Peaks
at
51,600
shares
on
13
May
Within
a
week
it's
below
1000
shares
per
day
(17
May)
Perfect
power
law
decay
–
no
spikes
aLer
launch
aLer
a
big
influencer
finds
it
belatedly
11.
12. 0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
15-Apr 22-Apr 29-Apr 06-May 13-May 20-May 27-May 03-Jun 10-Jun
ConPnuing
ripples
even
a
month
aLer
a
launch,
as
new
communiPes
and
community
influencers
discover
the
video
600
people
find
&
tweet/RT
the
video
on
15
April,
before
Dove
officially
tweet
it
(@Dove_Canada
on
16th)
Peaks
on
Day
3,
the
17
April.
Doesn't
show
the
rapid
power-‐law
decay
of
the
news-‐
driven
searches
Secondary
peaks
when
it
spreads
into
new
communiPes
&
is
noPced
by
new
influencers.
E.g.
@DoveUKI
on
19
Apr
26. 20
8
8
2
Dove Real Beauty!
Ryan Gosling!
Cmdr Hadfield!
Turkish protest!
Lifespan varies (continuous period at 500 shares/day)!
27. Although none of the variables alone
proved useful to identify a viral
phenomenon, all of them correlate around
two main models of viral spread
28. Spikers vs Growers!
High Volatility"
Fast to Peak
High Velocity
High Shareability
Shorter Lifespan
Lower Volatility"
Slower to Peak
Lower Velocity
Lower Shareability
Longer Lifespan
29. But what makes a meme spread along the
first or the second model?
30.
31. All the videos stimulated a similar higher than average
emotional reaction."
(52-56/100 Sensum Score / Based on GSR).
32. So can the audience composition instead explain why
memes develop along one of the other model?
33. 35
30
34
29
Dove Real Beauty!
Ryan Gosling!
Cmdr Hadfield!
Turkish protest!
All memes were similarly amplified
(average Visibility of a post containing the meme)!
35. Since both Amplification and Globality
seemed not to correlate with one or the
other model of virality we then looked at the
demographics engaged with each meme
36. 30 Years!
66%
34%
White!
Christian 55%!
Jewish36%!
!
Students 9%!
Journalists9%!
Web devs 8%!
Senior Managers 7%!
Musicians 6%!
!
@NASA!
@StephenFry!
@BarackObama!
@DalaiLama!
@Conan O’Brien!
!
Technology!
Science News!
Photography!
Music!
Comedy!
!
London 11%!
Toronto5%!
New York 3%!
Dublin 3%!
Vancouver 2%!
!
37. 19 Years!
21%
79%
White 81% !
Black!
Hispanic!
!
Christian 67%!
Muslim 24%!
!
Students 15%!
Sales 10%!
Journalists 4%!
Photographers!
Artists!
Stylists!
Admin Staff!
@KatyPerry!
@E.DeGeneres!
@TaylorSwift!
@JustinBieber!
@LadyGaga!
@KimKardashian!
!
Comedy!
Music!
Fashion!
TV/Film!
Health Issues!
Sports!
!
London 5%!
Toronto 5%!
New York 4%!
Riyadh 3%!
38. 26 Years!
50%
50%
White 99% !
Muslim 94%!
!
Students 12%!
Musicians 8%!
Senior Managers 8%!
Web Developers!
Journalists!
Engineers!
Graphic Designer!
Teachers!
@CemYilmaz!
@SertabErener!
@AbdullahGül!
@BarackObama!
@ConanO’Brien!
@WikiLeaks!
@Nytimes!
@BBCNews!
!
Politics!
News!
Tech!
Football!
Music!
!
Instanbul 50%!
Izmir 32%!
Ankara 4%!
Bursa 1%!
39. 18 Years!
26%
74%
White !
Black!
Hispanic!
!
Christian 84%!
Muslim 9%!
!
Students 33%!
Musicians 13%!
Actors 4%!
@JustinBieber!
@TaylorSwift!
@KatyPerry!
@MileyCyrus!
@DanielTosh!
@SnookiPolizzi!
!
Comedy!
Music!
Dating!
Extreme Sports!
!
NYC 6%!
London 3%!
Los Angeles 2%!
Chicago 2%!
40. As we couldn’t find any correlation between demographic
traits and virality models we then turned to the structure of
the audience by mapping the social graph (followers/
friends) of the people who shared the meme
50. But what is causing higher or lower
fragmentation within an audience?
51. 32, male, white, CAN/USA,
into science, tech and
comedy
30, male, white, UK, into
tech, comedy and music
32, female, white, USA/NYC,
marketing professional
16, female, white/hispanic, USA/
LA, into teen pop and reality tv
25, mixed, white, Turkey/Istanbul,
into politics, sports, web
21, mixed, white, Turkey/Izmir,
into politics, sports, web
17, female, white/black/
hispanic, USA/Texas, into
teen pop and reality tv
19, female, white, Global,
into comedy, music, tv
57. “Virality” is a relative concept depending on
the audience of reference
58. “Virality” is not just a property of the content, it’s also a
property of the audience.
Or as Jonah Peretti put it, Virality is 50% great content
and 50% distribution
59. Great content spreads fast or slow depending on the shape of
your audience and how you are leveraging it with your
distribution strategy
60. The audience you are trying to reach is fragmented into
sub-communities of age, profession, interest
61. Using network analysis you can identify these communities by
mapping the social graph of your target audience
62. The broader the appeal of your content the more fragmented
your audience is going to be
63. The more fragmented the audience,
the more targeted the distribution needs to be
64. Wide appeal = Grower = spend more on seeding strategy to
connect communities and sustain diffusion over time
Narrow appeal = Spiker = spend more on community
management to absorb + amplify impact
65. So if you want your content to go viral, don’t just put the video
out there and see what happens…
66. Study your target audience and plan your distribution strategy
based on a community-map,
not just on a list of
“influencers” (who might all be part of the same community)
But most of these variables seem to correlate
Based on these correlations, there seem to be two types of memes: Spikers and Growers
Spiker
Higher Volatility
Faster Peak
Higher Velocity
Higher Shareability
Shorter Lifespan
Grower
Lower Volatility
Slower Peak
Lower Velocity
Lower Shareability
Longer Lifespan
But what makes a meme spread in one or the other way?
All the videos we tracked stimulated a similar higher than average emotional reaction within the target audience (52-56/100 Sensum Score). sensum.co
sensum.co
First thing we looked at is amplification
Amplification is a measure of the average "visibility" of the meme.
Where there more influential people in the audience of one of the memes that helped it spread faster?
Lower for Turkish protests – primarily shared in Turkey, a slightly newer (though still very active) Twitter market
Lower on Ryan Gosling – perhaps as funny content appeals to a younger audience
Highest for Dove (video with a message) and Commander Hadfield (tapping into an older Bowie fan audience, plus big influencer RTs)
Mostly equivalent, so NO.
Is globality affecting speed? Does a more global or a more local meme spread faster?
The answer is again, no
So we took to Bayesian statistics to analyse the demographics of the audience
Does the demographics of the audience affect the way content goes viral?
There doesn’t seem to be any correlation with demographics as they are completely different
We run demographics analysis on the top clusters within each audience and identified higher demographic diversity in Gosling and Dove, and lower demographic diversity in Hadfield and Turksih
Virality doesn’t exist as such. It’s a relative concept, totally depending on the target audience
And this shouldn’t come as a surprise by now
Virality doesn’t exist as such. It’s a relative concept, totally depending on the target audience
Virality is 50% content and 50% distribution
Great content will spread fast or slow depending on the shape of your audience and how you are leveraging it with targeted distribution
Your audience is fragmented into sub communities based on social connections
Your audience is fragmented into sub communities based on social connections
The broader the appeal of your content the more fragmented your audience is going to be
The more fragmented your audience is the more effort you have to put on distribution
The more fragmented your audience is the more effort you have to put on distribution
Before launching your next campaign, map the communities in your target audience and identifies the hubs by community
Mapping the communities within your potential audience is key to find the right trigger, identify the right gatekeepers to escalate the diffusion and have the right strategy in place to support a spiking or a growing meme.