Case Study: Breaking Through to Low-Attention Voters

CHALLENGE: Many Moveable Voters Shut Down when Faced with Political Messages

One hallmark of the Trump era has been the constant barrage of discussion around him. Whether in the news, online, or in social settings, there has been no shortage of available information — much of it highly polarizing.

So when it came time to develop messaging for undecided voters, the first question was how on Earth anyone could not know what they thought about him. It certainly wasn’t because of lack of opportunity to learn, like one could have supposed in earlier eras with less pervasive political information.

This chart shows why reaching those who don’t follow political information is hard but critical for Democrats. Those who consume the least information are always the most persuadable, and they’ve also leaned heavily conservative in recent years.

Logically, then, one had to assume that if these voters didn’t know what they thought about Trump, they must be avoiding knowing. For whatever reason, when they see or hear political information, they put up mental barriers to keep it out. 

This left a very tricky challenge for 2024: how do you persuade a voter who refuses to hear what you have to say?

METHOD: Embed Political Information in Content that Already Interests Moveable Voters

To approach this challenge, I teamed up in 2024 with Movement Labs to launch a political blog — but not just any political blog. No, this political blog talked almost exclusively about college football.

Almost.

Into each football post, we inserted a small amount of political content. We hypothesized that the football content would keep male voters paying attention, while the political content would be just enough to persuade them.

For article topics, we selected stories that were timely and relevant to the contemporary news arc in the sport: the NCAA’s punishment of University of Kentucky, the questions around lightning rod head coach Deion Sanders, the formidable offense of the University of Georgia.

2024 Georgia Bulldogs quarterback Carson Beck

We combined these topics with political messaging that had recently tested well for others, largely drawn from the Doppler newsletter and Open Mic webinars. For example, we saw that talking about Kamala Harris expanding freedoms was powerful, so we worked that into a post.

For pretesting, we ran individual blog posts through the Grow Progress panel testing platform. For field testing, we promoted the posts through digital ads and surveyed the exposed populations. Then we looked at the results.

RESULTS: Enormous Movement in Pretesting, but Followup Needed

The findings from pretesting were absolutely incredible! The star performer was the post about the Georgia offense, which produced massive amounts of movement among just about every category of persuadable men you can think of.

That level of movement is virtually unheard of from this type of test. And it didn’t seem to be a small sample size fluke either, as the audience exposed to this treatment numbered over 1,000. (Sometimes people run these tests with as few as 250 subjects.)

Unfortunately, the field test we ended up running likely wasn’t the right fit for this type of intervention. When we exposed populations of men to ads promoting the blog posts, then surveyed those populations after a few weeks of exposure, the results were basically null.

But click through rates on this type of ad tend to be low, maybe 1-2%. The value in the ad usually comes from those who see it but don’t click it, absorbing its message nonetheless. Whereas in the case of a blog like this, you would expect (based on prestesting!) to observe a large impact on those who read the blog post and no impact on those who didn’t. With this test design, it’s unlikely enough of the audience read the posts to have moved the general population significantly.

Still, the magnitude of these results suggests there’s something powerful going on here. It would be worth looking for alternate ways to test this type of intervention in the field in order to better capture the impact it makes. If you would be interested in collaborating on further testing, let me know.