How Data Science helped win a Mayoral election
Facing a two-time incumbent in the client’s first race for higher office, the campaign team wanted to identify undecided voters and likely donors using advanced campaign analytics.
An A.I. model was developed using voter files, external data, conventional survey, donation history, and ProxyPoll data. ProxyPoll is a proprietary technique leveraging online behavior of registered voters to indirectly poll hundreds of additional voters with no direct constituent interaction.
Pairing voter files with behavioral data, the A.I. was able to identify the key attributes of a person that made them likely to vote for, donate to, and support a particular candidate. The key tasks assigned to the model included:
- Predicting a registered voter’s likelihood to show-up for the primary and general elections
- Predicting whether those likely voters were supporters, detractors, or still undecided
- Of the supporters, identifying those who had the highest likelihood to donate to the campaign
By aggregating all the campaign analytics into a centralized Voter Strategy Dashboard, the campaign team was able to quickly identify high likelihood to vote and undecided voters for strategic canvasing efforts, marketing campaigns, and donation requests. The dashboard further identified household of 3 or more undecided voters (1,600 in total) that the client was recommended to engage directly. Targeted canvasing was a highly efficient way to deploy limited resources and was a deciding factor in the Mayoral Candidate’s narrow victory.
By running the campaign using A.I. the campaign was noted as one of the most advanced and strategic campaigns in the state’s history.
Industry: Politics, Governmental Affairs
Skills: ProxyPoll, Propensity to Vote Modeling, Undecided Voter Classification Modeling, Propensity to Donate, and Voter Strategy Dashboard