Академический Документы
Профессиональный Документы
Культура Документы
PEER TO PEER
President How social networking took Obama to
the White House and how consumer
behaviour modelling can harness the
power of online social networks.
WhitePapers
America’s first peer-to-peer president
Introduction
2008 was an extraordinary year in US politics. A woman and an African-American
fought for what felt like an eternity for the Democratic nomination, with Barack Obama
finally triumphing not just as the Democratic candidate, but also as the 44th President
of the United States.
But 2008 was an extraordinary year for another reason: it was the year in which the
growing use of online social networks was recognised by the campaign teams and
turned into a formidable political marketing tool.
2008 was the year that social networking went into politics.
This working paper focuses first on the clear advantages garnered by the Obama cam-
2008 was the year
paign team through harnessing the power of online communities with the development
that social
of innovative online tools: tools that allowed Obama not only to harvest campaign
networking went
donations at an unprecedented rate, but also to identify and engage large sections
into politics.
of a previously hard to reach and disenfranchised electorate. The paper explores how
powerful messaging delivered to a myriad of online social groups, allowed unprece-
dented numbers of voters to find their voice and purpose within a campaign that pro-
mised to mend the perceived sense of mass-scale dislocation within US society.
Whilst the results of these efforts have been well celebrated since Obama’s win in Nov-
ember 2008, this paper then goes on to explore a less well documented perspective of The simulator
the campaign: how would it be possible to mimic such a compelling, decentralised and predicted a popular
amorphous process, whilst at the same time maintaining a degree of control? Would vote outcome across
wresting control from the communities destroy the very potency they promise? the battleground
states of 55.7%
Considering this dilemma – and presenting a solution – this paper then describes how Obama to
the proprietary Havas Media Consumer Behaviour Simulator was able to track mega- 44.3% McCain...
drivers for voter choice over various channels during the campaign and predict their demonstrating the
potency through until Election Day. As well as offering insights in to how various chan- precision of the
nels were assisting or hindering the candidates’ efforts, the Behaviour Simulator was tool, in what was
also able to accurately predict the popular vote result across the battleground states, such a complex,
thus delivering this sense of control and influence without overt disruption. The simula- dynamic and volatile
tor predicted a popular vote outcome across the battleground states of 55.7% Obama campaign.
to 44.3% McCain. The actual recorded result for those states was 54.3% to 44.1%,
demonstrating the precision of the tool, in what was such a complex, dynamic and
volatile campaign.
The authority of the Behaviour Simulator is further demonstrated when alternative sce-
narios were modelled, removing online channels from the campaign models. Without
these channels, Havas Media can demonstrate the results of the popular vote would
have been startlingly different.
© Havas Media
America’s first peer-to-peer president
It is an immutable fact that Obama represented the change the country confirmed it
was looking for. As an African-American from extremely humble origins and with the
most eclectic life story of any candidate, Obama embodied his key message from the
outset. But beyond this broad-appeal rhetoric, Obama and his team also recognised
another immutable fact: in order to reach, energise and empower as many voters as
possible, they needed to turn to the internet in a way previously unheard of in political
campaigning.
Obama’s use of the Internet and his team’s social networking strategy delivered the
Illinois Senator to the White House.
“We didn’t start with much money or many endorsements. Our campaign…was built
by working men and women who dug into what little savings they had to give five do-
llars, ten dollars and twenty dollars to this cause.”
© Havas Media
America’s first peer-to-peer president
Online, however, was not simply a more effective way to gather donations, but instead
was a compelling way to take the election to the voter. Donating became a social event
in itself with donors able to set personal targets and watch online funding thermome-
ters rise in line with their efforts – a potent source of fuel to drive further volunteering
efforts amongst an electorate increasingly hungry for change.
The Obama strategy enabled an unprecedented level of grass roots creativity and
ingenuity, allowing the campaign to revitalise and re-energise itself in ways that would His increasingly
naturally chime with the communities. One clear example of this is the iPhone appli- unassailable
cation that allowed users to mobilise their friends and contacts through the device in fundraising
key battleground states. The strategy also allowed for rapid and effective responses advantage allowed
to negative messaging and robo-calls in the battleground states (arguably producing Obama to push
strong results for online channel effectiveness). into Republican
heartlands with
The extraordinary success of MyBO also presented the Obama team with an opportuni- messaging, forcing
ty to command the more traditional channels of mass media advertising. Team Obama McCain to use
was able to flood traditional TV with commercials, spending an unprecedented $293mi- his resources in
llion on ads between January 1st 2007 and October 29th 20081 (McCain spent $132mi- increasingly
llion during the same period). defensive responses.
© Havas Media
America’s first peer-to-peer president
To provide answers to these questions, Havas Media Intelligence has been closely stu-
dying the effects of social media and has developed a proprietary model that effectively
simulates the behaviour of key constituents in any campaign across multiple media.
Using Agent Based Modelling techniques, this ‘Simulated Consumer Behaviour’ tool
allows organisations to accurately predict how brand communication strategies will
play out across these various channels.
Taking the view that the Obama/McCain campaigns represented the largest product
launches for 2008 if not beyond, Havas Media Intelligence decided to deploy the be-
haviour simulator in the context of the 2008 US presidential election campaign in order
to not only understand the relative effectiveness of various channels for both parties,
but to predict the popular vote outcome on November 5th.
Using Agent
Based Modelling
techniques,
this ‘Simulated
Consumer
Behaviour’ tool
allows organisations
to accurately
predict how brand
communication
strategies will play
out across these
various channels.
© Havas Media
America’s first peer-to-peer president
The qualitative focus group research allowed Havas Media Intelligence researchers to
Three principle
develop a deeper understanding of voters’ decision-making processes in a political
characteristics
context, to explore the right voter language and terminology and to start to build a
of the candidates
framework of the relevant touchpoints for voters. These focus groups were populated
were defined as
with Democrat, Republican and ‘Undecided’ voters (three distinct segments), with field
being mega-drivers
work completed in August 2008.
for voting behaviour:
From these focus groups, three principle characteristics of the candidates were defi- leadership,
ned as being mega-drivers for voting behaviour: leadership, consistency and trust. consistency
and trust.
With these insights, the second, quantitative phase of research was initiated. Talking
to 1212 voters across all three segments, this Havas Media ‘Connect’ study analysed
a number of characteristics: propensity to connect with other voters, the relevance of
each mega-driver for each segment, perception of each candidate in relation to each
mega-driver, as well as the relevance for voters in each segment, of various communi-
cation channels (TV, print, online, word-of-mouth etc).
The results of the ‘Connect’ study allowed researchers to define four principle channels
for voters that were considered highly influential: ‘word of mouth’ (physical one-to-
one conversations); ‘experience’ (such as witnessing debates, interviews and editorial
around the candidates, as well as local meetings); ‘online’ (blogs and social networking
sites) and ‘mass media advertising’ (paid messaging).
2. Battlefield states: those that were too tough to call for either candidate (New Hampshire, Vir-
ginia, Florida, Ohio, Michigan, Minnesota, Iowa, Missouri, Colorado and Nevada); those that were
‘McCain leaning’ (North Carolina, Georgia, West Virgina, Indiana, Louisiana, Arkansas, Montana);
and those that were ‘Obama leaning’ (Maine, Pennsylvania, New Jersey, Wisconsin, New Mexico,
Washington and Oregon).
© Havas Media
America’s first peer-to-peer president
The outputs from these qualitative and quantitative phases then provided the inputs for
the Consumer Simulator tool. As an agent-based model, the tool simulates consumer The tool predicted
behaviour through probabilistic analysis, recognising the propensity for virtual ‘agents’ a popular vote
to interact with the various defined media and the potential outcomes of those inte- majority for Obama
ractions. With such a dynamic and evolving campaign, it was important to continually within the
‘feed’ the simulation with updates on advertising investments from both parties, to battlefield states
ensure the simulation mirrored real-world influencing factors (something that would not of 55.7% to McCain’s
normally be so volatile – or reactive at least – in a normal planned media campaign). 44.3%. The recorded
result for those
Whilst the simulation ran from early August (as the first inputs became available from states was 54.3%
the qualitative research), it was not until towards the end of October that it was felt the to 44.1%.
evolving media investments were more or less played out; in other words, there were
no more significant spikes in spending to factor into the simulation. As a result, during
the last week of October, the tool predicted a popular vote majority for Obama within
the battlefield states of 55.7% to McCain’s 44.3%3. The recorded result for those sta-
tes was 54.3% to 44.1%.
60
50
40
30 Obama end
55.7%
McCain end
44.3%
20
10
3. For simplicity, no other parties were factored into the vote, i.e. the model assumed 100% of the
voting electorate would vote for either Obama or McCain.
© Havas Media
America’s first peer-to-peer president
The simulator tool also allows for detailed analysis of which mega-drivers - trust, lea-
dership and consistency - contributed most to this result, as well as which defined
channels proved to be most effective in communicating this mega-driver to voters.
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
start
start
start
start
start
start
start
start
end
end
end
end
end
end
end
end
0.1
0
Average Perception Trust Perception Consistency Perception Leadership Perception
As can be seen from Fig. 2, whilst both candidates enjoyed clear increases in positive
perception overall (block 1), the shift for Obama was more significant, except, interes-
tingly and somewhat counter-intuitively, in the case of Consistency (block 3).
Most striking of these shifts in perception towards these mega-drivers, is the Obama
shift in Leadership. With this mega-driver for voters, the simulator shows a shift from
broad indifference (0.38 approximately at the outset of the simulation) to an incredible
0.95 by the end of the simulation. This represents a near total agreement amongst
voters that Obama’s leadership qualities were superior.
We can see voter perception around this mega-driver for the candidates diverge from
mid-August in the Fig. 3, suggesting McCain may have enjoyed strong gains in this
attribute earlier in the campaign (considering he also ended the campaign with a net
increase in voter perception, despite the downturn during this period).
Having already defined the principle channels through which voters could perceive shifts
in these mega-drivers, the simulator allows for detailed analysis of how these channels
contributed to – or indeed detracted from – gains or losses in voter perception.
© Havas Media
America’s first peer-to-peer president
0.9
0.8
0.7
18Aug08 25Aug08 01Sep08 08Sep08 15Sep08 22Sep08 29Sep08 06Oct08 13Oct08 20Oct08 27Oct08 03...
Remaining with the Leadership driver, Fig. 4 shows clearly how the various channels
The ‘Experience’
influenced the overall shifts in voter perception for both candidates.
channel (debates,
There are two clear insights here: firstly the ‘Experience’ channel (debates, interviews, interviews, editorial,
editorial, meetings etc) proved to be vital for Obama. This channel represented in ex- meetings etc)
cess of 50% of the overall positive influence of all the channels for Obama, compared proved to be vital
to a small – but crucially negative – influence for McCain. Undoubtedly, the grass roots for Obama.
community activity so well documented around the Obama campaign is demonstrated
in this result. In addition, the effect of online channels (specifically not online editorial
but rather social networking sites, blogs etc) was less negative for Obama than for
McCain. Again, this result may well support the more efficient approach the Obama
team adopted in dealing with negative messaging (as detailed and linked to earlier in
this paper).
It is also interesting to note that Word of Mouth, whilst being a prolific channel in the
campaign (as shown in Fig. 5), failed to demonstrate any significant effectiveness in
influencing voter preference. This is almost certainly due to the fact that one-to-one
dialogue around the candidates tended to take place amongst like-minded individuals
and within environments where the propensity to change minds was almost zero.
© Havas Media
10
America’s first peer-to-peer president
50
40
30
20
10
-10
-20
-30
0.5
-0.5
18Jun08 30Jun08 14Jul08 28Jul08 11Aug08 25Aug08 08Sep08 22Sep08 06Oct08 20Oct08 03Nov08
It is also worth noting the significantly more positive effect the mass media channels
had on the McCain campaign, despite Obama’s impressive blitzing strategy. The rela-
tively recent historical dominance of mass media channels and McCain’s ability to use
these channels to such effect, prompted Havas Media Intelligence to consider an alter-
native outcome to the campaign: a campaign where the experience channel – driven
largely by online community engagement – had not been an option.
© Havas Media
11
America’s first peer-to-peer president
Re-running the simulation with the absence of this channel – both in terms of dissemi-
As consumers
nating information and allowing for voter-to-voter interaction – the simulation presen-
move evermore
ted a very different election result. McCain would have won 50.84% of the popular vote
into the digital
in the battlefield states.
domain, Havas Media
With the simulator’s ability to isolate the specific effectiveness of certain channels in Intelligence’s
such a nebulous and dynamic campaign, it is possible to show that in a more tradi- Consumer
tional campaign environment, McCain would have won. This puts beyond doubt the Behaviour Simulator
argument that MyBO delivered Obama to the White House. offers brands
and their owners,
opportunities to
explore dynamic,
effective and,
Conclusions above all, safe ways
to navigate the
In 2008, the US electorate voted for change on an unprecedented scale and in doing immensely powerful,
so, ushered in a new chapter in American history: America’s first black president. As but potentially
has been shown in this working paper, this journey for Barack Obama was made pos- turbulent waters
sible by his campaign team harnessing the power of the digital media; not just to raise of social media
campaign funds (although this was in itself an extraordinary feat) but to identify and networks.
engage huge swathes of the electorate that, until this campaign, had sat disinterested
on the fringes of US politics. In the digital media age, a new campaigning model was
born. To quote Joe Trippi, seasoned Democratic campaigner early in the campaign:
“We’re scratching the surface. We’re all excited because he’s got a million people
signed up – but we are 300 million people in this country. We are still at the infancy
stages of what social networking technologies are going to do, not just in our politics
but in everything. There will not be a campaign in 2012 that doesn’t try to build a social
network around it.”
To harness the enormous opportunities in digital, social media took a unique pers-
pective on organisational strategy. But for any organisation or brand considering at-
tempting to emulate the successes of Obama’s communication strategy, the ability
to identify and predict the effects of various social channels once the message is well
and truly beyond the boundaries of the organisation and in the hands of the consumer,
must be considered vital.
Havas Media Intelligence’s Consumer Behaviour Simulator has demonstrated with ac-
curacy, its ability to identify channel effectiveness and predict the outcomes of what
has been inarguably the most dynamic, hard-fought and volatile US presidential elec- For more information
tion in history. on Havas Media
Intelligence’s Consumer
As such, as consumers move evermore into the digital domain, Havas Media Intelligence’s Behaviour Simulator,
Consumer Behaviour Simulator offers brands and their owners, opportunities to explore please contact:
dynamic, effective and, above all, safe ways to navigate the immensely powerful, but Enrique Martin
EVP Havas Media,
potentially turbulent waters of social media networks and in doing so, build stronger, Research & Modelling
closer and more dynamic relationships with their consumers. enrique.martin@havasmedia.com
© Havas Media