Hal Malchow on Reinventing Political Advertising -
Hal Malchow is a seasoned political consultant and writer with a successful career in politics. As the chairman of MHSC partners, Hal was an innovator in control group experiments and advanced data analytics to improve political targeting and campaign strategy. He’s worked with the US Olympic Committee, Planned Parenthood, and five Democratic presidential nominees. Through his work modeling voter lists, he set many different fundraising records for the Democratic National Committee. He has been recognized for his revolutionary work with the Sisk Award in 2005 for reshaping the marketing industry and his induction into the American Association of Political Consultants Hall of Fame in 2016. His work is also chronicled in Sasha Issenberg’s book The Victory Lab: The Secret Science of Winning Campaigns, and has written political strategy, political fiction, political thriller, and young adult fantasy novels. We asked Hal 7 questions about his career, experiences, and his new book Reinventing Political Advertising.
You can find our How to Win a Campaign Podcast interview with Hal Machow here.
Your book talks about the ineffectiveness of political advertising and how to improve its effectiveness. How would you improve it?
One of the polls done recently by YouGov examined the percentage of voters who say they paid less attention to political ads versus all other ads, and this has grown from 28% to 44%. 58% say that political ads have no influence at all on their choice of who to vote for. Measuring persuasion advertising involves contacting a lot of people and is very expensive, but two political scientists, David Brockman, and Josh Kala, pulled together results from 46 experiments to determine the effectiveness of political advertising. They found that there were very significant effects on vote choice in primaries and ballot measures, but among general elections, the effect size was zero. The good news is you can fix this if people pay attention.
There are two issues; one is the targeting. For decades we have had one formula for deciding which voters we need to talk to, and that formula was which voters are undecided. It was targeted by looking at the continuum of the likelihood of being a Democrat or a Republican and targeting the people who are in the 50%. We have been asking the wrong question! The question is not whether someone is undecided; the question is “Can we move these voters with political advertising?”. The resounding answer is this: the people who are moved by advertising are low-information voters, the people who know very little about politics.
For example, in 2010, the New Mexico NEA did an experiment where they sent three mailings to their members and asked a control group of members who they were voting for to see whether the mailings they sent had moved any voters at all. The stats said they moved three points, which is a good result. But at the last minute, the survey company asked Todd Rogers, who is Executive Director at the Analyst Institute, “Todd, you’ve paid for one more question. What would you like to ask?”. Todd gave an answer that no one at the Analyst Institute or the NEA understood. He said, “Ask the voters whether or not they know which party controls Congress”.
Now, the universe here was people who voted in 2006 and 2008. Some people voted in all major elections. One-third of them could not tell you which party controlled Congress. Among the voters who knew which party controlled Congress, the movement was six-tenths of a percentage point in the direction of the Democratic nominee. Among the people who did not know, the movement was nineteen points, thirty-one times the movement of those who actually knew which party controlled Congress. Why would advertising work in the primaries and in ballot measures, but not in the general election? The reason is that people know a lot more about general election candidates than they do about the other two.
The second reason that advertising is working so poorly is that television advertisers are not paying any attention to the data, and they're not testing this stuff. We have pre-testing that goes on now, and in a lot of ways it will help you understand whether an ad is better than average. But it won't tell you in any sense what the actual effect size is because the voters who look at these ads are paid to look at them, and usually people are not paid to watch ads. People hate ads, people pay money not to see ads. Advertising is a weak way to communicate with anyone, and political advertising is worse.
Does low-information voting vary based on the type of race?
Yeah, absolutely! Three academics accessed election data and television advertising for 4,500 races over a ten-year period. Well, what did they find? They found that if you had a one-hundred-advertisement advantage in a presidential race, you move the needle two-hundredths of a percentage point. If you were in a Senate race, you moved four hundredths. If you were in a congressional race, you moved eight hundredths. If you were in an Attorney General's race, you moved sixteen hundredths. If you were in the state treasurer's race, you moved a third of a percentage point. Well, why is that? Most people know a lot about the candidates who are running for president, but no one even knows who's running for state treasurer. Again, this underscores the importance of understanding which voters are uninformed.
We conducted an experiment where three different mailers were sent during the Pennsylvania 18 special election. We sent regular mailers, false positive mail with controversial statements that looked as if they were sent by opponents, and impartial informational-only mailers. Neither the traditional mail nor false positive mail had any effect on voters, but the information-only mailing moved voters a point and a half. Although you may think a point and a half is pretty small, that’s big when you compare it with other measurements across the board; in all mediums, a point and a half is a big deal.
So, they rolled the mailers out; I think they sent 80 million pieces of mail in the congressional, gubernatorial, and Senate races. After analyzing all of the precinct returns across the country, it was determined that we would not have won the House in 2018 had we not used the tactic of information-only mailers. Why did this work? It worked because it looked like credible information; people don’t know much about candidates or politics, particularly as you move down the ballot, and the challenge is getting them to believe what you say. Polls show that 58% of voters don’t even count advertising at all in their choice of candidates. The credibility that this format provided was a breakthrough! Have you ever seen a TV ad try to do this? The degree to which practitioners in our profession have ignored data is shocking.
In your book, you talk about social pressure and being much more deliberate about targeting. As folks utilize some of these tactics, like social pressure, do you see diminishing returns on some of the tactics that do work?
I haven’t been on top of the GOTV mail. The two big turnout tactics are social pressure and vote-by-mail, and we’ve been able to fine-tune all of this. The reason we can fine-tune it is because it's so cheap to measure, right? You take your treatment and your control, match it to the voter list, and see who voted; you don’t have to make a single call so the expense is nominal. That’s why we have progressed so much and have so much hard knowledge about what works in voter turnout, but in persuasion? No one is spending the money.
A prognosticator predicted that in 2024 there would be $16 billion spent on political advertising, making it the tenth biggest advertising market in the world. Do you think people in the commercial area who are doing this advertising at large volumes aren't measuring their effects? Of course, their effects are easier. It's a lot more complicated for us. But if we put some of that 16 million towards gaining knowledge and learning new lessons that would cause us to be able to formulate even better strategies. Businesses, specifically in the technology industry, can spend 24% of their revenues on research and design, right? We could spend a half percentage point. You take 46 experiments on how advertising affected general elections, put them together, and realize you didn't affect them one bit. There is a crisis in political advertising and it's as much a knowledge and targeting crisis. Until we solve this, we are going to have to continue experiments that tell us we are not making any difference at all.
What advice do you have for organizations, candidates, and consultants apart from actually spending a part of their budget on testing?
The number one idea is to pick up the telephone and call Catalyst or Target Smart and say, “When are you going to get us a voter information model?”. This is critical to taking these lessons and putting them into effect. Take the New Mexico experiment; what if, in every race where we had a contested race between Democrats and Republicans, we were able to increase advertising effectiveness a mere four times? How many more races would we win? We would win a lot of them! But we are not going to win them until practitioners pay attention to research.
Some people only pay attention to national politics and not to local politics at all. How do you solve for that variable?
If you are doing state legislative races, then probably the voter information dimension is close to useless because people don’t have much information about these candidates. If you build a model that attempts to measure the information level of each voter, you might have 10% that you could drop, and it may make sense to do news programming. Now that we know this about low-information voters, do you know where our television advertising has been going? The percentage of television ads placed on news programming has been 55%. If you took that ad spending and placed it randomly, without any strategy whatsoever, you would probably double our effect sizes because we are focusing all our ads on the people who won’t be moved.
So the question is “Who can you persuade?” right? Well, you have this low information variable, which is shockingly powerful in terms of predicting who would move. Another variable that has developed in recent years is the medium of the ad; some people can’t even be reached on broadcast television. Some spend ten hours a day on the internet and some don’t read their mail. I know Catalyst has excellent measurements on the predictions of these variables. Unsurprisingly, tests have shown that Internet ads work a lot better on a voter who spends ten hours a day on the Internet! If they watch a lot of TV, television ads are more effective.
You create a persuasion score combining the information level, the media coverage, and being in the middle of the partisan scale. This can create three different indexes: a persuasion index for digital, a persuasion index for television, and a persuasion index for mail. Once you have three scores, you can multiply them by turnout probabilities so you’re measuring how likely it is that voters are persuadable and showing up to the polls on election day. This would simplify the campaign manager’s job when making these decisions, and it would make everything much more effective. If you could get four times the effect size from low-information targeting and digital targeting, it would be substantial enough to reshape the outcomes of a lot of elections.
What advice do you have for folks currently going into politics and advocacy work?
Don't dwell on the conventional wisdom, you know, read my book and think about how to do these things differently.
You’ve never been worried about going against the direction of others, how do you keep that internal compass and create your own new ideas?
Well, one, I don't believe anything just because someone tells me or because everyone's doing it. Two, I don't sit around and wring my hands about decisions; I look at a problem, I make a decision, and then I figure, “How do I actuate this?”. Then, I do it. I also think I cut down the time between contemplation and action just by my personality.
You’ve created many successful businesses and built great teams of people working together toward a common goal. Any advice on doing this? Asking for a friend!
It's fairly simple: you treat your employees really well. You give them opportunities for advancement, and you focus on getting engaged with new ideas and training to improve their skills. I once went six years without a professional-level staff leaving the firm, and that’s probably made me prouder than just about anything. And Joe, it was a sad day when we left you, but you’re better suited for running your own company!
You were involved in the data side working on data projects and the creative side working on direct mail designs. How do you keep the ideas coming?
Just by being absorbed in the work! I get the most pleasure from having an idea that goes against all the conventional wisdom and making it happen, measuring it, and being successful. If I'm in the business and I'm doing the same thing everyone else is doing, am I changing anything? If they use another consultant, they're going to get the same results. My only opportunity to change the outcomes is by doing things differently.
Do you have a favorite book, movie, or podcast that you draw inspiration from when writing your books and building these businesses?
I went back to school and took three statistics courses to try to get myself ready to do bigger things. And I read a lot! There are a lot of books I love, but in terms of helping me professionally, the best book I ever read was David Allen’s Getting Things Done. Before I read this book, I always had a big pile of paper on my desk and things unorganized, but he taught me discipline and basic organizational components. I remember after I read the book I organized all my paperwork and labeled files, and a staff person walked into my office and almost fainted! It helped me a lot, and it made me 50% more efficient in everything I did.
Talk about the evolution of your career journey-
Well, I started in fundraising. This is important because in fundraising you get a report card on everything you send out. No one ever goes to the conference room and says, “Let's have a meeting to decide whether this worked”. It's all right there in front of you, and I was just baffled because we were putting out all these pieces of mail and what everyone loves is the big splashy, dramatic piece. Does that work? I don't know. No one had any idea. So, I got interested in that measurement, and I funded the start of the Analyst Institute which I'm always very proud of. The targeting was somewhat a piece of that because you could use statistics to look at different mediums and how they worked for turnout. Finally, I just got kind of disgusted with the campaigns we were doing. This is 2010. Issues like climate change are confronting the American people. We could all fry. We're losing social security, and Medicare is heading for insolvency. We've got crime problems, we've got deficits, the list of problems was staggering. And what were we doing mail about? The main pieces that we did that year were either about our opponent taking a pay raise or us not taking a pay raise. It was about the trivial stuff because these other issues were too big, and the candidates feared them. I found it disgusting. So that's when I got out. At that point, I spent about eight productive years with the Voter Participation Center helping guide their experimental agenda. A lot of what we learned are important elements of everything we do, particularly in turnout.
Thank you Hal for all your work. You have been a pioneer and mentor in moving political advertising, digital, direct mail, and testing forward.
Listen to the How to Win a Campaign Podcast interview with Hal Malchow here.
Have questions about political advertising? Drop us a note.