July 27, 2016 | Eric Bradlow & Colleen O'Neill
Customer Analytics & Voting Behavior
The Penn Wharton Public Policy Initiative and WCAI recently co-hosted an event that discussed the opportunities that exist when voting analytics and marketing analytics combine their methods and data.
Listen to the recap
This episode recaps the half-day event co-hosted by Penn Wharton Public Policy Initiative, the Wharton Customer Analytics Initiative, and the Penn Program on Opinion Research and Election Studies and featured premier Wharton Faculty.
Read the recap below
Imagine a Venn Diagram with two circles. The left hand circle is labeled, “Voter Analytics,” and the right hand circle is labeled, “Marketing Analytics.” While the amount of separation and overlap between these two circles depends on who you ask, it is quite clear, however, that significant overlap exists based on the discussion at a recent event hosted by The Wharton Customer Analytics Initiative and the Penn Wharton Public Policy Initiative.
In the business world, Marketers have access to CRM data that’s linked to transactions that give them behavioral metrics over the course of years. Businesses can study and experiment with this data to understand and predict what their customers will do in the future.
In the voting analytics world, data-based strategists have records of every registered voter in the U.S. This dataset contains information like age, gender, location, voting history (not who but when), and party affiliation; not to mention some public information like census data and some purchased credit data from FICO. Campaign strategists can study and experiment with this data to understand and predict what their customers voters will do in the future.
What would happen if a dataset existed which linked voter information, with CRM information, with transactional data (for example, someone’s Kindle inventory), and potentially some internet search history? Then, further imagine that one applied the same advanced analytical methods used in both worlds.
This super-charged dataset could potentially provide a highly accurate picture of who’s in what state of the “(Voter) Buyer’s Journey.” From this point, targeting different voter segments is almost exactly the same process as targeting potential customers.
If campaigns put this dataset to work, run experiments, and apply the models, it’s well within the realm of possibility that targeting efforts could move 2-3% of the vote from one candidate to the other. Following that, campaigns could target the swing voters, just like companies target swing buyers. Even moving a modest 2-3% of the vote is enough to completely change the outcome of an election and eliminate a lot of guesswork.
Now imagine an election season that applies rigorous analytics! This could save a democracy a huge, year-long speculative forecasting season every four years.