AI is increasingly important to all industries, including political fundraising, as Sterling Data Company found.
There will be more than 1,000 elections in the United States in 2022 at the state and higher level. And as of June 30, 2022, six fundraising committees associated with the Democratic and Republican parties have reported raising a combined $ 1.3 billion. That’s a lot of political campaigns and money.
Raising and spending that money effectively for campaigns is where specialist firms like Sterling Data Company enter the game. Sterling is a national Democratic political data firm focused on fundraising. Whatever your political preferences, Sterling’s use of artificial intelligence is instructive for pretty much any organization looking to gain competitive advantage.
Low code, big donors
The fact that Sterling increasingly depends on AI to drive its business isn’t particularly surprising: Who isn’t using AI today? But what is surprising is how Sterling Data uses AI. For starters, they don’t hire data scientists and AI specialists. Instead, using Akkio, a SaaS no-code AI platform, they simply upload Excel spreadsheets into the cloud. Despite the simplicity of this approach, Sterling’s AI-driven capabilities are on par with much larger – and more expensive-to-hire – firms boasting dedicated teams of data scientists and AI experts.
So how does it work?
SEE: Artificial Intelligence Ethics Policy (TechRepublic Premium)
Sterling works with candidates from the city council level up to congressional races. Among more than 1,000 clients, they consult for Beto O’Rourke’s race in the Texas gubernatorial campaign. To support their work, Sterling has amassed a database of more than 30 million campaign donors over the years, with each donor defined by as many as 500 different variables like how much given, voting record, age, magazine subscriptions and more.
According to Sterling CEO Martin Kurucz, each congressional district in the US averages up to as high as 50,000 donors with the mean around 17,000.
“How do candidates find people who will be interested in their cause?” The chicken asked. “That’s where we come in. We have a ton of variables, and we try to figure out who the most probable people are. “
The data analytics component is just a part of the extraordinarily complex challenges of developing fundraising strategies for candidates, Kurucz said. There exist many different analytic models for different campaign scenarios, such as whether a politician is a long-shot candidate in any given race.
To show how AI can profoundly impact the result, Kurucz walked me through a hypothetical race for a congressional candidate in Minnesota. The candidate provides data they have gathered themselves. From that list, joined with Sterling’s own database, and given the budget the campaign has to spend fundraising, Sterling needs to find out who’s most likely to donate.
Working with more than 500 variables per name, Kurucz gives Akkio parameters for a model he wants to generate that will surface the most likely donors. Depending on the dataset size, Akkio returns results in 30 seconds to 30 minutes.
“I haven’t found anything else that can do that, not even close,” Kurucz said. “The resulting model is unique to that candidate. Now you can deploy it to predict ‘zero,’ not a donation, or ‘one,’ a likely donation. And we test it and revise the model. Best of all, you can back-test it. “
The models created by Akkio perform up to 400% better than models created the usual ways, he said. His best models historically showed 100% + ROI within three months for candidates. Akkio models can be ROI positive within a month.
That’s the power of AI. But the magic is that a no-code approach puts this power in the hands of a daily practitioner who no longer needs expensive data scientists to achieve these results.
Kurucz says he can create and run his AI models on a plane on his laptop.
“Spending on list analytics in political campaigns is booming like crazy,” he said. “But only a few companies have all the different puzzle pieces to make it work – now more and more AI-powered. So that’s where most of this is headed. “
Disclosure: I work for MongoDB but the views expressed herein are mine.