Last night was the first Democratic presidential debate and almost immediately afterward the “winners and losers” articles started appearing. But how do we know who’s really winning?
Right now, polls indicate that several of the Democratic candidates would beat Trump if the election were held today. But polling famously got it wrong in 2016. And as pundits, reporters and political consultants seek to assess the state of the race they are increasingly likely to supplement polling data with other types of information, such as search, social engagement and, especially, location data.
Search data useful but ambiguous. Search volume and query data can indicate a number of things about popular interest in candidates. While useful, there are also some problems with relying on search data alone, such as ambiguity inherent in the data and the often uncertain relationship of queries to subsequent behavior.
Search volumes may also not be entirely representative of the voting public. This is similar to the way that online social media activity around brand sentiment doesn’t reliably predict offline buying behavior.
Location history and event attendance. Location data and offline movement tracking is a newer and potentially more accurate indicator of intent and future voting behavior. Location analytics firm Gravy was able to (fairly) reliably predict the outcome of the 2016 New Hampshire presidential primary using voter attendance at political rallies and candidate meetings across the state. The company extrapolated outcomes from “which events were most heavily attended and for how long.”
Other data such as regular church-going or attendance at gun shows or events promoting the environment can similarly be used to identify and predict voting preferences. And there’s lots of existing data about brand affinities, shopping preferences and political affiliation.
Targeting voters based on offline shopping behavior. Gravy CEO Jeff White explained that not all location data, like search query data, is revealing of political preferences but that by layering data and combining various sets of visitation patterns it can get much closer. “The campaigns have profiles of the voters they want to reach,” he told me. Gravy (and its competitors) can identify those groups using mobile ad IDs but in a privacy compliant way — he was careful to emphasize. “They share their taxonomy and we find those people.”
Event attendance, store visitation, change of life events data (children, job changes, new home ownership), which can all be tracked using location, can be combined to build reliable and predictive models. That data can then be used for political ad targeting and for predictive analytics: who’s likely to vote and for whom.
Why we should care. Depending on your perspective, this is either exciting or frightening. Regardless, offline movements and visitation patterns — which stores, business locations or events people go to — is very search like in revealing intent. A person consistently visiting car dealerships or open houses is 99% likely to be an in-market buyer.
Different demographic groups shop at Walmart vs. Nordstrom. People who regularly eat at Chick-fil-A have different characteristics (and potentially values) than those who are Taco Bell loyalists. In other words, consumer activation principles using location data can equally be applied to targeting and activating voters — and predicting election outcomes.
