By DAVE CARNEY, POLITICO MAGAZINE
Greg Abbott’s campaign can teach the 2016 GOP how to win.
As Greg Abbott takes the reins as the 48th Governor of Texas, political operatives on the right should take a closer look at his campaign and learn a lesson or two for 2016.
The 2014 election cycle was a very good year for Republican candidates across the country, but the challenge of re-claiming the White House in 2016 is staring us directly in the face. And while Attorney General Abbott’s 20+ point win in the gubernatorial election seems like a foregone conclusion in hindsight, it was not always a sure thing. In fact, at the point that State Senator Wendy Davis entered the race, she was the darling of national Democrats, as evidenced by the fact that much of President Obama’s top campaign talent descended upon Texas to help her turn the state blue.
Turning the Democrat dream of a blue Texas into the nightmare of a massive loss happened because we ran a campaign that used every tool and strategy a modern campaign has at its disposal, and did so in the most efficient and effective way possible.
Too often campaigns try to fight the last winning war. For Texas Democrats that meant trying to remake their campaign in the shape of Barack Obama’s successful 2012 re-election. And we saw the results.
Candidates running in 2016 should see this as a cautionary tale. The Obama campaign did many good things, but every campaign is different and candidates need to adjust to changing demographics, technology and public opinion.
With that in mind there are certain principles on which successful campaigns like ours can build. We were guided by three basic principles that every Republican running for President needs to apply to their campaign: (1) talk to one audience; (2) measure outputs, not inputs; and (3) test and retest.
Let me explain:
Talk to one audience. Many campaigns today take the first step and bring on the right mix of talent from the likes of polling, predictive analytics, digital marketing, media buying/optimization, data management and field operatives, but allow them to operate in silos.
Television ads are bought based on Nielsen ratings. Direct mail is sent to voters with certain “modeled attributes.” Online ads target voters based on their interests and online activity. And the field team knocks on doors of voters based on their past vote history. None of these are necessarily bad ideas or tactics, but everyone is talking to a different audience.
Campaigns need to settle on a base-scoring system using predictive analytics and dynamic modeling that will serve as the common currency for all aspects of the campaign. It is this currency that defines the audience to whom paid media (television, online, radio), digital marketing and the campaign’s grassroots/field organization are communicating.
The Abbott campaign settled on parameters based on their likelihood to support Abbott and their propensity to turnout to vote early in the campaign that produced a universe of approximately five million voters. Everything the campaign did communicated to them or a subset of them.
We targeted voters whose attributes were most similar to those of our supporters, which we were continually refining based on data across platforms from various sources: online, donors, volunteers and activists. Using ongoing modeling we were able to score and update every voter in the state as supporters, persuadable targets or lost causes. This involved tracking and collecting every contact the campaign made. This allowed us to identify a large pool of persuadable women and Hispanic voters, both of whom we blew the doors off of on Election Day, as the exits polls revealed. The exit polls showed Greg Abbott won 54 percent of women, 50 percent of Hispanic men and won 44 percent of Hispanics overall—all of which are traditionally strong Democratic groups.
All field contacts, all our online advertising and even our television used set-top box data to target these voters.
Once we had identified our universe of five million voters using predictive analytics, we matched these audiences online. We then ran audience-specific messages to this targeted persuasion audience using pre-roll, expandable video and Facebook promoted video ads. Over the course of the final six months, we were able to reach 74.43 percent of our target persuasion audiences online and served them an average of 22.51 ads each at an extremely cost effective average of $0.34 per voter.
Measure outputs, not inputs. Too much time is wasted worrying about motion without caring if there is progress or not. We look at how many impressions an online ad receives, how many gross rating points are behind a television ad, or how many calls a phone bank makes.
None of these matter.
Campaigns need to look at how many favorable voters we identify, how many targets an ad persuades and how many voters an email turns out to vote. The purpose of the campaign is not to talk to the most people, it is to mobilize the most votes. That is how success should be measured.
Which volunteer is more valuable? The one rushed through their calls and made 200 dials in a night or the one who only talked to 50 voters but convinced them all to get out and vote. By standard campaign metrics volunteer one would get a gold star and volunteer two would be “the guy who talks and talks to everyone he calls.”
Our campaign never asked field staff to make a certain number of calls or knock on a certain number of doors. In fact, our field team had the freedom to do their job as they saw fit. All we did was require them to identify 250 Abbott supporters every week. Some of them spent a lot of time knocking doors themselves while others organized volunteers and never talked directly to voters at all. All that mattered was delivering results.
At the end of the campaign our field staff were able to identify more than 1.9 million pro-Abbott voters. That is more than 70 percent of the voters who ultimately turned out in support of our campaign (and would have been enough to win the election). In nearly forty years of working on campaigns I’ve never been involved in a statewide campaign that was able to ID more than a third of their voters.
Test and retest. There shouldn’t be any sacred cows. Most campaign consultants think they know what works. They’ve been sending that get-out-to-vote mail piece in every election cycle since 1982 and 58 percent of their clients have won because of it.
The reality is that media habits and technology are changing so fast that no one knows what works today or more importantly what will work tomorrow. Large campaigns with multiple media markets have the ability to test tactics in near real time and make adjustments accordingly.
In 2014, campaigns spent more than 60 percent of their media budgets advertising on just 20 shows even though there were more than 1,500 that they could have selected. This may have worked back in 2004 when we were saying goodbye to Friends, but even with population growth of more than 23 million, the most watched show today would not even crack the top 15 in 2004.
When it came to television placement we did truly turned the world upside down and upset many Texas television stations used to attracting large buys from Republican candidates. We ignored traditional Nielsen ratings, instead creating custom ratings for our target universe based on our five million targets.
When compared to the kind of buy that would have been made using Nielsen ratings as a guide, our custom buy in just the Houston market produced a savings of more than $80,000 per week while fully saturating our target audience with our message. That translates to more than $4 million statewide over the course of the campaign, while delivering an additional 29,000 combined rating points to our target audience across every media market we advertised in—in other words, our average target voter would have seen our ads an extra 21 times.
One of the biggest surprises of this change in buying came when we looked at network distribution. Traditionally, Republicans spend more money on CBS than any other network. Their popular primetime shows drive a lot of Republican spending, but we found was that our target audience just wasn’t there. Instead we spent more on NBC, CW and independent stations that we typically would have.
Running a statewide campaign this way in 2014 was very difficult and the challenges facing the presidential field are beyond anything that we have previously seen. Testing and analysis leads to more decisions and more analysis than we have grown accustomed to and will require significant focus on the development of decision-making infrastructure (people) on campaigns. A truly data-driven campaign requires accepting the stories data tells and allowing data to inform decisions, even if it is at odds with gut feelings or the way things have always been done.
That’s the most important lesson for consultants and strategists approaching 2016: The way things have always been done won’t be the way the things are done in this upcoming election cycle. We fell behind the smart use of these new tools in 2012 and now have to apply the lessons we learned in 2014 if we’re going to stand a chance of winning back the White House.
Dave Carney is a New Hampshire political consultant who has worked in Texas politics since 1993 and was a top advisor to Rick Perry and Greg Abbott.