Welcome to episode #91 of The PPC Show where we interview the best and brightest in paid marketing. This week we're joined by Louis Belpaire, Director of Paid Media and Analytics at Silverback Strategies, to talk about running Facebook political ads and the success he had with them.
Tune in to learn:
- The details on running political ads on Facebook post Cambridge Analytica
- His three part Facebook ad creative and audience targeting strategy
- Results from the campaign
Listen to Episode
Louis Belpaire oversees all aspects of Silverback's paid media relationships and leads a team of specialists who manage paid search, social, display and other advertising programs.
Prior to joining Silverback Strategies, he worked on the online marketing team at Groupon, helping scale and optimize SEM campaigns in 40+ countries and develop innovative tools for campaign management.
Louis is an avid triathlete, an Ironman finisher and a huge fan of cycling events.
Show Notes and Transcript
- Here's the link to the blog mentioned in the episode
- Also listen to Merry Morud episode where she talk's about Cambridge Analytica
JD Prater: Louis, welcome to The PPC Show.
Louis Belpaire: Thanks for having me, JD.
JD Prater: Yeah. Louis and I connected. He got this really amazing Facebook ads for political campaigns case study that we're going to be talking about, which I am super pumped on to really dive into a lot of this. Especially within the last quarter, when you think about some of these changes. But before we do, Louis, tell us a little bit about yourself and where you're working, and what you're up to these days.
Louis Belpaire: Yeah. So I'm the Director of Paid Media and Analytics at Silverback Strategies. We're a full-service performance marketing agency. We have around 35 employees, and we're located right in the heart of Old Town in Alexandria, Virginia. So we're right outside of DC. We essentially have a paid media and an SEO team, a content marketing team as well with in-house designers and video resources, and we also offer analytics and development services.
JD Prater: Oh, nice. Very cool. Very cool to see the full stack there to helping people out on their digital strategies, so cool. That was Silverback, so go check them out out there in the DC metro area. But before, let's dive into this man. So Louis and I connected, and he was telling me about this really cool campaign that they ran on Facebook. And so we're gonna be kinda walking through his process, really through his strategy. And then we're gonna insert some really cool stuff of like what's happened, and what are some of those challenges that came up. So we could really start off with Cambridge Analytica happened, right? We all remember this. This was right in the middle of March, so kind of start there with this man. Whenever you think about Cambridge Analytica and you think about running political ads, tell me that mindset.
Running political ads on Facebook post Cambridge Analytica
Louis Belpaire: Yes, so this was a little bit crazy. As you were saying, this is maybe 30, 60 days after the Cambridge Analytica scandal. And essentially we have a candidate for the democratic mayoral primary here in Alexandria and his campaign manager that came up to us as an agency. And they were looking for an agency that was local, so that was pretty easy. But also, they wanted to go big on digital because they knew that his core audience was people in their 30s and 40s. And so that you have to go digital. And Facebook would probably be a big part of the play. So with that in mind, I thought with the timing of things, this was almost too interesting of an opportunity to let go of. So I really wanted to jump on it. And there's also the fact that as a candidate, we thought that his position on some of the local issues were pretty well aligned with our philosophy as a business. So we decided to give it a go. But the timing was definitely a little bit crazy. Because Facebook was changing a lot of things when it comes to political around that time. So there were definitely a few challenges there. Shall I get into a few of these details?
JD Prater: Yeah. Yeah, let's go into it. I'm sure everyone's wanting and eager to know. What are some of these constraints? And what happened as a result? And what's still possible?
Louis Belpaire: Cool, yeah. So I think the main hurdle would be the authorization process that advertisers ... Now we know it's a thing, right? But as of a month and a half ago when I started the campaign, I essentially started running ... and we'll get into the detail of the strategy I think a little bit down the road. But when I started launching the campaign, just a few days later I remember it's a Tuesday morning. I get into the office and my Gmail, Facebook notifications have blown up. And basically all of my ads have gotten disapproved. And what happened as we know now is that Facebook was starting to use their AI to identify ads with political content and automatically blocking them unless you're an approved political advertiser. So what that means is that agencies and people who put ads on Facebook have to go through a three-step process, submit their official security information, get a piece of mail with a code and basically verify that they're a US resident.
Facebook blocked ads with political content
And so the timing of that wasn't great. Because for a solid 48 hours, I wasn't sure how I would get my ads re-enabled. Luckily, we do a little bit of work here in DC with nonprofits. So I have a few people on my team who had to go through the same process. So I was able to use their login to re-upload my ads. But I think that's the main big thing. They're monitoring very closely the types of ads that are put on Facebook obviously because of what happened. And people have to get authorized. So I think that's the first main thing.
The second one would be in terms of sharing content related to news. So our candidate maybe three or four weeks into the campaign, right when we were in the heart of things, got endorsed by The Washington Post. So as you can imagine, being in the DC area, that's a pretty big deal. They don't often endorse local candidates. So we're like, "Yeah, we're going to really promote that and get it in front of people who didn't read the article yet."
The issue with that is unless you have verified access to a website on Facebook ... Unless you own the domain, especially when it comes to news, you can't just put an ad on an article and then use that as your ad format. So we kind of had to use a creative workaround there where we showed a picture of the candidate with sort of a Washington Post endorsed logo. But we couldn't direct people directly to the article. So that's kind of one of the new limitations.
In terms of what you can do though, you can still target people based on their political views, based on the fact that they're likely to engage with political content on Facebook. And you can still pretty much target custom audiences, right? You don't have as much visibility to the estimated reach. But you can still do that. So when it comes to political advertising, there's a lot of databases out there of past voter data. It doesn't tell you who someone voted for because that's obviously private information. But you can see who voted and where on what list. So essentially, you can re-target voters from past elections. Which is kind of really important for these local elections where turnout is usually pretty low. And that's a good way to make sure that you get a little bit more exposure.
JD Prater: Yeah. I mean, what's crazy is if you were to start this in February. February, you would have had custom audience estimates, you would have known how many people, right? So then you would have been able to run these types of political content. Cambridge Analytica happens in March. You start advertising in April. And now all the things that you've been so accustomed to have just changed. So absolutely crazy. Because now we're going to get into the actual campaign strategy, right? So now you're living in a post-Cambridge Analytica era. And you still have to drum up awareness. Get the vote out and get people out there to actually vote, but also vote for your guy. So let's kind of talk to about how you executed this strategy. Let's ... yeah. Walk us through kind of your threefold strategy that you implemented.
Louis Belpaire: Absolutely. So I think we were lucky in this case. And it's not always the case with clients that they had done a lot of great research beforehand. So there's actually a poll that was conducted in Q1 this year by a campaign manager where they interviewed about 400 local residents. And they had a very precise view on what was going on here in Alexandria. So we knew essentially three things. We knew that the opponent, the current mayor had much more recognition, especially in some areas of the city. She was projected to win and get about 53-54% of votes because of having more awareness in the city.
The second thing is that they were very specific demographic groups that have different interests in the election, right? So for example, older demographic were more likely to oppose new development. Parents of young children, they have specific issues at heart regarding the education budget and then things like that. So we knew it would be at some point really important to have a targeted message in front of a very specific audience. And obviously Facebook lets you do that very well.
And then finally I was talking about turnout a little bit. So turnout there, we knew that turnout would be critical. That's pretty much how you win a local election, whether your base shows or doesn't. So knowing that, we kind of created this full frontal strategy on Facebook. And the other thing I didn't mention is, it's a pretty small election. I mean, there's 160,000 people here. The candidate did not have a super comprehensive website. So sending people to a site was kind of the question. So this is really a time where we had to run a campaign from start to finish within the Facebook ecosystem. Which is kind of exciting. Because with clients, especially as a performance agency, very often it's about getting link clicks, getting people to the site so that they can convert and buy. But here it's completely different. It's more of an awareness play. The conversion is getting a vote. So it's a very different ...
So in terms of how we structure that is essentially, six weeks prior to the campaign, we started this awareness phase, right? So we're picking the awareness objective. The goal here is to optimize based on ad recall. And we're essentially showing people a video ad that's a very pretty quick ad with some background on the candidate. And we're also showing a cannabis ad. And I think the cannabis ad that's on the recording was really what functioned the best. So the cannabis ad really allowed us to create sort of a competitive ad. I'm sure you've seen those on TV. But that's not quite what we're trying to do. We're just really trying to expose the current mayor's voting record vs. our candidate's policy as quoted by local publications. The Washington Post really showed their facts to people side by side. And the cannabis is such an engaging experience that it did that very well.
So I was talking about our ad recall on video where that was 16 to 17%. With the cannabis we had more success there because we saw that on average people would see more than 60% on the cannabis. And it was pretty lengthy. So that's pretty good. And time-wise, they would spend more than 30 seconds within the ad unit. So why try to send people to a website where you know that only 50% will actually load it, and then a third of that might read through it, when you can do all of that within the Facebook experience? So obviously it requires good design resources. But something that's worked really well for that awareness phase.
JD Prater: Yeah. I gotta jump in on this part. Because this is crazy to me. So I love this idea of using Facebook as full frontal approach. And you're almost just like, "Don't even go to the website. Just stay here. I'm going to help you out. I'm going to give you content that is made for Facebook consumption." And so you start out with some video ads. Love this cannabis ad idea though. I mean, fantastic idea of going through. And 30 seconds of cannabis? That's nuts. It's crazy to think about how much time that is. And I like how you compared it to a website. But even just think about your own Facebook experience and how much time that you are spending within the actual Facebook platform. I've seen stats that are ranging 40 minutes a day. You think 40 minutes a day, and you just spent half a minute engaging with just my one ad. That's a win too. So hats off to you guys too, for thinking through those types of ad units, rolling those out. Especially in that awareness phase, but also getting those results. So let's move on to that second part of it. So that was the awareness phase you said, for ... that was six weeks before the election, correct?
Louis Belpaire: Yeah, that was six weeks to about four to three weeks prior to the election. So really trying to warm up the audience. And I think we got in front of 70,000 uniques at that point. So the majority of that 120,000 voters potentially saw that. Yeah. So the second phase, the hyper-local messaging. That's where we sort of concentrated about a third of the budget. So here we're not optimizing on grant awareness anymore. Not trying to get ad recalls as much as we're trying to get people to engage, right? So our video team actually produced a bunch of one-minute video that were endorsements from not local celebrities, but local public figures, I would say. So those were hyper targeted.
For example, women over 60 plus saw a testimonial from the candidate's mom. That was where she was actually talking about his upbringing and how she tried to have some specific values through his childhood and then how that makes him a good candidate for becoming the next mayor. So that's one audience. And then parents of young kids saw a specific message that's related to education. So we saw a lot of engagement there.
And one difference with, I would say for example the commerce campaigns that I'm used to be running is that this isn't as much of a sort of set it and forget it. Set it and monitor your KPIs because there's going to be a lot of comments that come through. Specifically since this is related to politics. So this is where the kudos really go to the candidate. Because I knew after hours, he was logging into his Facebook and really making a point of answering almost every comment. People would just jump in and say, "Well, this is great, this one-minute video thing. But what's your position on this particular issue?" And he would chime in there. So we got a lot of conversation started in that stage. And although that was just a third of the budget, this is where most of the reactions, comments, shares and even page likes, that's where we got all of that.
And the cost per engagement I think was around 15, 20 cents. So similar to our cost per ad recall at the stage before that. And yeah, that ran for about two to three weeks. Basically up until 48 until election day.
JD Prater: Nice, nice. That's really awesome, man. Just kind of thinking through. So you're doing some ads. You're doing video. We're doing Canvas. And now we're ... Are you guys retargeting those audiences, or were you just ... I mean, the audience was small enough as it is. You're just kind of targeting the same people again.
Louis Belpaire: So we really ... We leverage retargeting. But we leverage that really at the very end, in the last 48 hours. So for the last two days, we did do that. People who engaged with the candidate, who had seen the campaign video or the ad local videos as well as his page fans. Because that's obviously the strong base. We added these as basically an audience that we targeted in the last 48 hours.
And there's something a little crazy that happened 24 hours before. So as I'm sure you know, the Washington Capitals, the Caps won the Stanley Cup. So there was going to be a big parade in DC on the National Mall on Tuesday. I think that was June 12th. And I guess it's not so much a big deal, except that our core audience is those 30-something and young parents. And there's a ton of Washington Caps fan in Alexandria. So the campaign manager was sort of really worried about that. So very last second I also put an ad set for Alexandrians who are fans of either hockey or the Washington Caps saying, "Enjoy the parade today but make sure you stop by the voting booth on the way to the nation's capital." So I'm sure that that didn't hurt.
But yeah, those last 48 hours were pretty important in terms of just making sure we were leveraging the success from these top and middle frontal campaigns and reengaging people at the last moment.
JD Prater: Oh man, that's fantastic. So last two hours. Or sorry, last two days. We're talking that last 48 hours. Caps win, fantastic. You're really psyched but it's like, "No! Come back, make sure you vote!" And so when you think about getting that message out, what were some of those tactics that ... maybe even ad units that you used again as you were kind of talking about ... I loved the ad copy example. Anything else there that you guys really kind of were focused on?
Louis Belpaire: So I think the key there was to use the reach objective. Because it's just about getting it in front of as many people that day. We're no longer looking for recall or engagement. It's about having the entire base show up. But in terms of messaging, we also went with something more simple where we're essentially boosting the post of the candidate that day, his organic posts where he's just telling people to get out and vote. And it's as simple as that. There's no need anymore for some very creative message or anything like that. It's about having that one-to-one conversation where the message really came from him.
JD Prater: Fantastic. And so what happened? What were the results? Tell us the outcome here.
Louis Belpaire: So it was really exciting day on June 12th. Because I didn't know this beforehand. But local elections really happen like national ones. So at 12 PM and 4 PM, you already get ... You basically get ... You don't get voting data, but you get turnout data. So they were already releasing turnout by precinct. And I had data from the previous election, and we also had data about his expected share of votes in each of these precincts. So essentially, we were able to project the number of positive votes he got, both at 12 and 4 PM. And I quickly saw, especially at 4 PM, that not only did the precincts that are very favorable to him got 15% more vote than in 2015. But that the ones that aren't were trailing a little bit behind. So there was a lot more people that showed up to vote.
And at the end of the day at 9 PM, they released the official number. And he won and got 53% of the overall vote. And what's great too is that the election mobilized I think I want to say around 20,000 people, where it was closer to 12,000 three years ago. So there was pretty massive turnout, all things considered, for such a local election.
JD Prater: Yeah. I mean, that's really big. I mean because you're also influencing people that maybe not be voting for your candidate, right? I mean, you probably got other people to go out and vote, whether they voted for yours or not. But I mean, that's really cool to kind of see that power really more than anything else of what you can do with really great ads and really great targeting, and when you mix those together, and how you can kind of mobilize people. It's a little bit scary. We're thinking about 2016. But at the same time, it worked on an election of this caliber. So when you kind of think about this, and this is still fresh. We're talking just a couple weeks ago. What were some of your key learnings that you want to pass along to advertisers that are in the same boat as you, as they might be thinking about their own elections as we kind of think. Mid-term elections are coming up this fall.
Louis Belpaire: Yeah. So I think the main learnings are around ... the fact that I'm honestly ... My background is in search marketing. And I have this very granular optimization background. And we're seeing over time that automation is sort of taking over, and you've had a few podcasts on that. I think one of my main takeaway is how fantastic the Facebook platform is. That once you set a goal, just optimizing towards that and taking care of this at delivery part. So I would say it's ... Once you get the concept of sort of using this sort of frontal strategy where as you get closer to the election you move from a more generic message to engaging people towards a conversation, and getting out the vote types of messaging. Once you've understood that, it's really about focusing on the quality of creative. Because eventually, that's what's going to make the difference.
And then I would also say that one differentiator is first-party data. So our candidate, one thing they were doing as they were canvassing, so going door knocking. They were collecting people's email, name and address so that we could reuse that to retarget them as custom audiences. Especially when they were favorable to the candidate. So there's still ways to kind of have an edge on the competition. And some of that is making sure that you have good first-party data. And that apply even with small, local campaigns like that.
JD Prater: Yeah. So as you kind of think about the future. I guess my two questions for you: would you take on another political candidate? And then the second part of that is, what advice would you give someone taking on a political candidate?
Louis Belpaire: So would I ever do this again? It's a little bit different, the types of clients we work with. And I think by definition, this might be better suited for PR firms that focused on political work in general. I think the fact that it was local and then there was a good fit, we were just curious to do it, is why we did it. But in terms of advice, I mean you have to be ready to work not around the clock. But this is very different than a typical campaign management process where ... I mean, campaign managers are used to emailing on the weekend. And candidates campaign on the weekend. So it's kind of a 24/7 thing. So it's A, you have to be ready to work around the clock for these types of campaign. And you have to be ready to jump in, especially as a candidate. And then put out fires.
JD Prater: Well, cool. Well man, it sounds like it went fairly well, though. I mean, you got a candidate that came to you. You accepted him as a client, which is not your typical client at Silverback whatsoever. I mean, you are not an agency that focuses on politics or taking them on. And then really great job on the execution of strategy. I really liked how you kind of walked us through, how you used the budget, what types of ads you saw. And then ultimately, the outcome of that. So I think it was also hats off to you guys for a really smart strategy, and good work at putting all that together.
Louis Belpaire: Thank you. Awesome.
JD Prater: All right, man. So let's wrap up. I'm sure you've got some people that might be interested in learning a little bit more about political campaigns. Or maybe they want to reach out and ask you some questions. Where can people find you, and how can they connect with Silverback?
Louis Belpaire: We're going to have a blog actually come out sometime next week, I believe. Our content team is working on that currently. So if you want to find out some more details on the strategy and the outcome, you can find that at SilverbackStrategies.com under our blog section. And then don't hesitate to reach out to me on LinkedIn or on Twitter. My handle is @b_louis.
JD Prater: Nice work on getting that one. And I'll make sure to link to that article in the show notes. So if you're going through and you're wanting quick access to that, I'll make sure to include that as well. So Louis man, thanks again for coming on and talking to us about some Facebook political ads. And I'm sure I know I had very curious and was wondering what's happening with them and what is going on, how is it impacting advertisers. But it seems like for the most part, it was fairly smooth sailing even though there was a lot of stuff that was in flux. So good work, man. I really appreciate it, and for coming on and sharing all of your learnings with us.
Louis Belpaire: Thanks for having me. I was definitely excited to get a chance to get on your podcast. Now I know I can tell my team that they definitely have to subscribe and listen to it.
JD Prater: All right. And I'll make sure to send you that $5 for that part. No. But thank you again for the shout-out. And for any of those of you guys listening, I mean feel free to ... if you go to blog.adstage.io, you can subscribe to our newsletter there. We'll make sure to include the podcast on that newsletter that comes out nearly every Thursday. And you'll be able to keep up with the podcast. And then also you can go subscribe. We're on Apple Podcast. We just got on Spotify, so go check us out on Spotify. We're pretty psyched there for you Spotify listeners to go and follow us there. And we'll be able to keep everything there. But until next week, thanks, Louis, for coming on and talking with us. And we'll see you guys all next week.