Obama Campaign Manager Jim Messina Talks Big Data at the Milken Institute's 2013 Global Conference

Strategic Foundation of the 2008-2012 Obama Campaigns

• The campaign adopted a "personalized campaign" model on the first day, utilizing data to inform nearly every major decision.
• The primary objective was to move and persuade individual voters based on specific behavioral insights rather than treating them as aggregate numbers.
• A "singular goal" was established to secure 332 electoral votes, a target successfully achieved.
• The data strategy evolved from 12 team members in 2008 to 165 in 2012 to support expanded data collection and modeling.
• The campaign built behavioral models to rank every American voter from 1 to 100 based on three key metrics: likelihood of supporting Obama, probability of voting, and propensity to be a ticket-splitter.
• Instead of the standard 800-person poll sample, the campaign conducted 10,000-person samples nightly for 14 months, enabling 62,000 daily computer simulations.
• This simulation infrastructure guided the allocation of a $1 billion budget, replacing traditional "smoke-filled room" decision-making with data-driven directives.
• The campaign achieved extreme predictive accuracy, predicting the final vote in Florida within 0.05% and missing no state by more than 0.5%.

Operational Tactics and Field Execution

• In Wisconsin during the final week, the campaign utilized six distinct iPhone scripts for volunteers, assigning specific scripts based on real-time voter data.
• Post-interaction, visitors received a personalized website (barackobama.com/[voter name]) loaded with their specific data points.
• The campaign developed a "targeted sharing" mechanism on Facebook that leveraged social proof to influence undecided voters.
• Over the final six days, six million users accessed a 20-second Michelle Obama video featuring five of their own undecided friends and neighbors; 78% of these targeted users voted for Obama.
• This strategy allowed the Obama campaign to secure a majority of undecided voters for the first time since Richard Nixon.
• The campaign implemented rigorous A/B testing across 20 million email subscribers, testing 24 versions of each email (including button size, font, sender, and imagery) to maximize returns by 82%.

Specific Revenue and Engagement Case Studies

• A "Dinner with Barack" fundraising event generated approximately $3 million per occurrence; adding George Clooney to the event tripled revenue to $12 million.
• Data analysis revealed that celebrity popularity (Clooney being "hot") drove the revenue spike, rather than the combination of candidates.

Future Outlook and Industry Trends

• The Republican Party conducted a post-mortem admitting a 3-4 year gap in data sophistication compared to the Obama campaign, with similar historical precedents set by Ken Melman's 2004 operation.
• Jim Messina predicts competitors will eventually replicate these capabilities, noting that the "sustainable advantage" relies on the sheer volume of data and time required to build sophisticated prediction models.
• The campaign emphasizes that modern data analytics (A/B testing, behavioral modeling) are now standard across tech platforms (Facebook, Google, LinkedIn), creating a baseline requirement for political organizations.
• The 2012 campaign team spent four years building the underlying data infrastructure, with the analytics director appointed on November 6, 2008.

Post-Campaign Strategy and 2016 Implications

• Messina notes the legal complexity of commercializing campaign data, as political campaigns cannot legally sell products without violating contribution limits.
• The team is exploring non-partisan vehicles to distribute data tools to nonprofit advocacy groups and issue-based organizations.
• For the 2016 election cycle, Messina warns that campaigns must "reinvent" their interaction strategies rather than replicating previous methods, stating, "if you run the same campaign you'll get beat."
• The Obama team aims to remain agnostic regarding which candidates use the data, intending to provide a "head start" to both parties to ensure the field remains competitive.