Are you getting the most out of your digital persuasion campaign?
Political targeting has grown leaps and bounds in the past few years with new innovations seemingly happening every day.
One of the biggest developments is the use of artificial intelligence (AI) to drive a campaign’s digital analytics, optimization, and audience identification.
What can AI Analytics do for your campaign?
We’ve identified five different ways that AI can be used to optimize your campaign in real time.
1. Deliver the right messages with the right messengers
They’re the age-old questions in political communication:
What is the message? Who is the best messenger?
Traditionally, those questions have been determined based on polling, ad tests, and other qualitative research (if a campaign is lucky enough to a larger research budget).
Today you can supplement a research budget of any size with real time statistics and recommendations driven by AI that tell you what messages work with certain voters and what messengers are the best to deliver those messages.
In the No on 2A campaign in Pueblo Colorado (pictured above) we utilized 57 unique creative executions at any given time. Through AI we were able to learn what messages and messengers voters responded to best – and make adjustments to creative delivery in real time. Our efforts showed dramatic overall video completion rates (71.37%) and engagement rates (35.22%). Our AI-powered digital and social advertising campaign was critical component to our overall victory on the ballot measure.
2. Creative Optimization: Make sure more voters watch your ads to completion and engage at higher rates with your content.
Have you ever wondered if images, colors, fonts, and creative presentations matter when it comes to click through rates?
But the answer can be very different among voters. AI-powered optimizations can look at traditional voter groupings such as party affiliation, geography, voting history, demographics but it can go far behind that by analyzing thousands of attributes – everything from travel habits to magazine subscriptions to vehicle ownership and beyond. AI can find patterns in these attributes where none seemingly exist and help you deliver advertising to best generate engagement among your targeted voters.
AI-Powered tech helped Eric Genrich win big in a seat that had been held by a Republican mayor for the last 16 years.
In Eric Genrich’s successful campaign to become Green Bay’s next mayor we were running over 100 pieces of display, video, and native creative at any given time on the web. Using AI-powered optimizations we tested images, colors, messages and ad sizes with every possible voter group we could imagine.
AI optimizations tested all the creative against each other and made a determination as to what was best to achieve high click through rates. For example, we were able to achieve a 1.5% click through rate among male Independents over the age of 60 with a purple ad with a black border. The industry benchmark for a similar banner ad is .08% – our results were 18.75 times that benchmark.
Our efforts on the Genrich’s winning campaign were recognized by the American Association of Political Consultants as the “Best Use of Data Analytics/ Machine Learning.”
3. Find new voter audiences receptive to your campaign message.
AI-powered technology allows a campaign to analyze thousands of attributes from individual voters who are interacting the most with your campaign ads. AI can find patterns in these attributes where none seemingly exist. From that analysis, a campaign may be able to create a new audience of voters not initially targeted but would be highly responsive to the campaign’s message based on the newly identified attributes. We’ve used this very effectively in several campaign. For example:
• No on 2A in Pueblo Colorado. We utilized AI-powered technology to identify and analyze key voter attributes within our original target audience that engaged with our creative the most. AI identified one audience stood out: people who prioritize “Low Cost Energy”. We were able to target an additional 6,000 people that fit that profile that we hadn’t previously fit our targeted voter profile. We expanded to include this new audience and found it to be incredibly perceptive to our message. This new audience watched our videos at a 72.87% completion rate and clicked our content (CTR) at a rate of 0.16% – much higher than industry standards (and in some cases higher than our other targeted audiences).
4. Exclude voters who are less responsive
AI-powered technology can also help campaigns to analyze thousands of attributes from voters who are responding the LEAST to the campaign’s advertising.
Once the “least responsive voters” are identified, you can shift digital resources to either new voter audiences or existing voter audiences that are performing better to increase your frequency of communication. For example:
• Yes on Proposition 19 in California. We were running ads to an incredibly well-modeled list. With a limited budget, we knew that we would have to look within these lists and “drop” the voters who weren’t engaging as much in order to increase the reach and frequency to the voters that were more responsive.
AI-powered technology showed us that voters who purchased luxury cars and young voters with lower incomes responded particularly poorly. Those exclusions allowed us to expand to a new voter audiences that AI identified as likely to respond well – this included pet-owning voters who like to be outdoors and voters who liked to read about technology and science.
By making the aforementioned exclusion of voters we were able to expand to a new voter audience that would have otherwise not received communications from the campaign. In the end the new audience had a significantly high video completion rate at 80.45% and a click through rate of .12% – higher than the industry benchmark.
5. Define an undefined audience.
In some instances campaigns start with a very large audience of voters as a target (for example – targeting every voting-age male in Michigan). AI-powered technology can help campaigns determine who is most responsive within that larger pool of votes. This allows you to zero in your ad buying dollars to reach voters with a frequency they will remember vs spreading your ad buy so thin that no one sees or remembers it.
In Michigan we were running a public service announcement educational campaign to a broadly defined audience making up roughly 32% of the state.
AI-driven reporting showed us that voters who didn’t like to travel and voters who rented their homes engaged with our advertising at a rate 4-6x higher than other targets. Based on this learning, we were able to adjust our focus to ensure a higher frequency to those voters most responsive to our messages. In the end, the campaign finished with a very high video completion rate (71.89%) and a CTR of 0.17%.