We had a torrid 90 day affair with political advertising. Here’s what we learned.
by Deacon Webster, Walrus CCO
In the three months leading up to the 2024 presidential election, we had an opportunity to do something that’s rare for an agency that works on brands: to create some political ads. This is typically an exercise left to the political experts who have mastered the art of making LOTS of ads fast and on the cheap. They have developed very specific mechanisms for determining what to say and whether it’s working that are quite different from what a Fortune 500 CPG marketer might be utilizing.
Our client was a small super PAC. They had no media money but they had a partnership with NowThis, a social media property with a huge following. They also had a big rolodex filled with production and Hollywood people who were willing to donate their time.
We typically dwell in the realm of paid content, not organic, but, given the lack of funds, the only way our work was going to be seen was if it achieved some degree of virality on social media. As anyone who has tried to make something “go viral” will attest, it’s not easy. You need to seed it and pray. To increase our odds, we wanted as many at-bats as we could get.
We ended up with 18 executions featuring the likes of Chris Parnell and Jeff Ross, hitting on a variety of issues from healthcare, to reproductive rights, to the environment.
The plan was to post them from our new, zero follower, handle on TikTok and Instagram and collaborate (i.e. co-post) with NowThis and the celebs in the spots that had big followings. We also had PR help with outreach to politically active celebs who might be willing to share a story or repost.
Over the course of the 30 day flight, the videos were viewed 16 million times, and shared 332,000 times. Four of the 18 videos we made achieved 1 million+ views.
Here are some things we learned along the way:
On Testing:
We had the chance to run a few of the campaigns through the testing apparatus at Future Forward which is the Dems’ biggest super PAC, and has build a finely tuned testing machine.
You have to test the actual finished spots in order to get an accurate read on effectiveness. All that stuff we regularly use to test brand advertising – story boards, animatics, board-o-matics – the political world learned long ago that those will not tell you anything. It’s got to be the actual ad. Intuitively this should not come as a shock. Production value and execution are extremely important. Interesting though that political advertisers have determined that there is zero insight to be gleaned from rough executions.
Length is not that relevant. We tested a few versions of a spot with different edits. The view time varied considerably from spot to spot. On one version people lasted six or seven seconds, on another 18 seconds. When we looked more closely at the dropoff points, in all versions it was when it became obvious that it was a political ad. Best practices had told us that people would only engage for a few seconds with any ad-like political content on social, so we should jam as much info as possible in the front of the spots before they skipped. In reality, length was not an issue – it was the message itself, or the way we were framing it. In the end we had multiple ads that were longer than 60 seconds all with high completion rates.
On TikTok and Instagram:
The algorithms are fickle. Many of the spots opened with actual footage of Donald Trump saying something that set up the concept that followed. Wondering if that was off-putting to our audience, or was getting our content throttled because it was political (which was happening on Instagram a lot in October), we posted versions sans Trump clip. In one instance we reposted a spot we had put up before but without Trump up front. The Trump version had only 8k views, but the new one shot to 2.7 million. Thinking we had cracked the code, we took a Trump clip off the next video we posted, and it only got 7k views. Meanwhile another one from that same series with a Trump clip hit 3.9 million views.
Conclusion: the TikTok algorithm is unpredictable. It’s good to have a lot of content.
Something that blows up on Instagram is not guaranteed to blow up on TikTok and vice versa. Our Airplane video was our most viral on Instagram – it got over 3.8m views. On TikTok – 472k. (Side note: someone put that one up on Twitter and it generated 4.2 million views before the platform took it down for no explained reason …) The healthcare spot that garnered 3.9 million views on TikTok got 231k on Instagram.
TikTok and Instagram are very different in terms of the way content moves through the platforms. At the moment TikTok generates more discovery than Instagram. Our collab posts on TikTok generally achieved higher numbers, probably because people on the platform spend more time looking at “for you” content than “following”. Instagram is the opposite. Follower content is prioritized by the interface while discoverable content is buried deeper in the experience. When Alicia Keys with her 26 million followers posted one of our videos to Instagram, it generated 900k views almost immediately.
On Instagram a shared story is far less impactful than a repost. For anyone looking to get some added reach via an influencer, you’ll get a lot more bang for your buck if you get them to post your content in-feed. Stories disappear after a day and only feature a fraction of the content – requiring an extra click to see more. In-feed posts live forever on the wall, generate far more views, AND feature the content in its entirety.
Don’t post on Friday. In a possible bright spot for humanity, people seem to avoid their phones on Friday nights, and thus everything we ever put up on a Friday was seen by only two or three presumably introverted people.
