This year I got to do serious political campaign work for the first time. Yes, I’ve been deep in the world of digital political advocacy for more than two decades, and I’ve helped out on small campaigns in the past, but this election cycle immersed me in the mechanics of working to elect battleground candidates for real. Stressful, maddening, mind-numbing, sleep-depriving — and more fun than anything else I can remember doing professionally. Two months of running as fast as you can, using every brain cell you can summon or borrow, constantly on the edge of brutal failure? Let’s do it again tomorrow, and I’ll buy the beer.
Not surprisingly, I learned a few things both about the mechanics of digital politics and about the bigger picture of getting people elected. Happily, my ebook held up well when confronted with hard reality, but this year’s experience taught me two solid lessons I’ll share today, one technical and the other perhaps anthropological.
Political Targeting is Personal
When my advocacy clients advertise online, they usually target people who meet certain criteria without knowing who those folks actually are. For instance, Google search ads naturally connect with people with an interest in the topic — otherwise, they wouldn’t be looking for information about it. On Facebook, a nonprofit might try to hit people based on demographics or location as well as interests, hoping to grow a supporter list or spark an online action. Only when a group targets its own supporters (or a purchased or borrowed list) via a “custom audience” does it know exactly whom it’s going to reach before they take the bait.
The electoral world works in the opposite way, to an extent I didn’t realize until I actually started running ads at scale. Unless a campaign doesn’t have the skills or resources, by default their digital advertising targets individual voters. For example, this fall I helped two Democratic state parties run video ads designed to support state legislative candidates. In both cases, I worked with voter files that ranged from a few thousand to almost a million names, turning them into custom audiences to upload to Facebook or a programmatic video ad system (DSPolitical). More on that endeavor soon, but the critical point is that we aimed our ads at individual voters chosen by the state parties’ data experts and matched by the ad platforms to those people’s individual online identities.
We supplemented this granular outreach with broader demographic targeting when possible, since no data model or matching mechanism is perfect, but overall we tried to reach specific people with content designed for their specific legislative districts. I’d gotten a hint that this was standard practice last year from colleagues frustrated at Google’s restrictions on audience-targeting on its properties (sample reaction from a data guy: “YouTube basically f—-d us this year”), but I didn’t realize how much voter-targeting was de rigeur in the campaign world until I actually did it on the ground.
Why are campaigns so specific in whom they target? Partly, cost. Why spend money on someone who’s not registered to vote, or who’s so clearly on the other side that you can’t possibly persuade them, or who’s almost certainly going to turn out and vote for your gal anyway? Also, think about blowback. Running ads at the other side might just persuade someone to turn out AGAINST you out of spite or frustration, a factor I suspect may have worked against some Democrats after this year’s TV advertising frenzy. Lesson learned, and I’ll be updating the book accordingly.
Even the Simplest Thing….
Pulling back from the nuts and bolts of campaigning, I learned something bigger: it’s almost a miracle that any of this stuff actually gets done at all. Let’s turn to Clausewitz for insight.
Everything is very simple in war, but the simplest thing is difficult. These difficulties accumulate and produce a friction, which no man can imagine exactly who has not seen war.
The same idea applies to political campaigning. On one project this fall, my colleagues and I had a pot of cash that wanted to be spent and an advertising platform that was eager to spend it. The trick? To get the money from one place to another without violating campaign finance rules, without breaking any of the contracts involved, and without having an intermediary siphon part of it off for their own purposes. We figured out the details in the end, but the process was maddening and far more convoluted than it needed to be. We wasted hours — too many of them, particularly when you count the number of people involved.
Other times, a project might be stymied by someone’s lack of a clear understanding of the rules, or a misinterpretation of an email or phone call, or simply by someone’s honest mistake — too often mine. Frequently the hardest part was getting the right person to finally say “yes”, since political organizations have to juggle so many priorities at crunch time that staff can’t possibly focus on everything at once. “Friction” indeed, and a good reason why political professionals tend to be risk-averse. If you don’t have time to think through all of the implications of an action, political and legal, the natural course is to put it off until you do. Particularly when the consequence of making the wrong choice could be an opponent’s attack ad — or a campaign-finance lawsuit.
Multiply each of those critical decisions by a few hundred or a few thousand and you get a sense how things that should be easy can turn out to be so hard to pull off in practice, particularly when no one’s had enough sleep. Good planning helps, but to steal the idea behind another military adage, no plan survives contact with reality. In the campaign world, simply pulling off what you hoped to do can be a victory in itself.
Next time, I’ll be more prepared but the challenges will remain — I suspect they’re inherent in a process where real power is at stake and many eyes are watching. But believe me, I’m up for another game. Want to join? You know where to start. And as always, let me know if you’d like some help getting on the field.