Google Ad Spending in the 2020 Presidential Race

Over the summer, I worked at a non-profit in DC exploring digital political advertising databases. While these companies have offered voluntary transparency measures, there’s little accountability and regulation in online digital spending.

Here’s what presidential candidates have been spending on Google ads, since Google made its data public in late May 2018.

2020 Democratic Primary Spending on Google Ads

The most illuminating things here are Tom Steyer and then Michael Bloomberg skyrocketing after they enter the race.

2020 Presidential Campaign Spending on Google

Google also offers some insights into which search terms advertisers spend the most on. For example, Beto O’Rourke’s campaign might have specified in their ad buys that they wanted their ads to come up when someone searched his name on Google or one of its platforms. What’s interesting about this data, however, is we don’t know who was spending on each keyword, just that there was money spent. Donald Trump’s campaign, for example, could be spending to place ads trashing Beto. There’s unfortunately no obvious way to identify who spent on which key word.

Search Term Spending over time

As states compete for relevance in the presidential primary, they move their primaries earlier and earlier. While Google ad spending is not specific to the presidential race, this map can give us at look at where digital ad spending is the most aggressive. Darker states are where the most dollars per capita have been spent.
Per Capita, Google Ad Spending by State

My next goal with the mapping may be to create a chronological animation, so we can see which states have an influx of digital ad spending as the primaries progress.

How to do this

• First, thanks to Michael Toth, for this blog post that I used to create the racing bar charts.
• Here are a few sources of digital advertising data:
  • Google
  • Facebook
  • Twitter
  • Snap
• Flourish also has a racing bar chart template
• I used the tidycensus package to create the geojson for the map in R, then created a style on Mapbox.

Limitations

• Lack of ad content (photo and video formats).
• Struggles with formatting details (date objects, color palettes, alignment, fonts)

Next steps

• Ways to identify and track dark money spending.
• Health care spending in California
• California-specific analysis, using congressional-district-level data.
• Twitter, Facebook and Snap
• Ad content