#INNDEM2017 Speaker: Michael Babyak


Name: Michael Babyak

Topic: Trump Campaign



Michael Babyak is the former Director of Marketing Technology at the RNC where he worked during the 2016 Presidential election cycle. He is now Co-founder and CEO of ArrowIQ, a cutting-edge digital, data, and technology company specializing in building self-sustaining digital ecosystems and automating the processes of growing and engaging audiences through personalized content.



He will share his experience building a winning, state-of-the-art Presidential digital fundraising operation from the ground up with the RNC and the Donald J. Trump for President campaign. He assembled and led the RNC’s Performance Optimization and Experiments Team (POET) that ran over 300 tests on DonaldJTrump.com content, generating over $30 million in added revenue.

Summary of Michael’s presentation

30 November 2017


Michael was part of the Republican National Committee (RNC) Performance, Optimization & Experiments Team (POET). His role was to help integrate the RNC and Trumps’ digital campaign teams. Mainly content testing via A/B testing. This enabled the team to understand their audience, serve them content which will push them to action, increase the chances they will donate and the donation amount, and finally, to go to the ballot and vote for Trump.


There are many tools out there which are built to help people do just that, to increase engagement. Optimally they want to use them for “good”, obviously there are a lot of bad ways to use technology but it is a choice everyone makes. Michaels’ team believes they used it for good, to increase engagement, and get people to take action.


Michael repeatedly mentioned that all of this is not revolutionary, it is simply ‘best practice’ in the digital marketing world. Politics is simply the last frontier for technology, in terms of the best practices people use for marketing. For some reason, politics is always a bit behind of the technological progress. Furthermore, if we are talking about tools such as A/B testing, which any of the major news sites use, as well as any of the major brands like Amazon, it was far from what the political sphere was used to.


Here are only few of the tools which are available, and which we used:
Adobe Target, Adobe Suite, Google Optimizer and Optimizely.


The POET team did have some advantages over other candidate teams:


  • They received complete trust and buy-in they received from their ‘principals’
    from the start, which allowed them to really institutionalise testing into the
    organisational culture of the whole campaign team.
  • The velocity of the whole operation and being able to start from scratch with a major brand (Trump).
  • The Trump team was radically transparent with what they were doing, they communicated internally very effectively, kept all the stakeholders engaged, educated them about what was actually happening, encouraged them to give feedback, and to participate in the process.
  • This was done through storytelling. This was really unique compared to other marketing efforts of other organisations on Facebook and Instagram.


Normally as a national committee for a party, you are required to build infrastructure such as an email lists, or basically a ‘well’ of people for the nominee to tap into and find the “right” people, to raise donations, and build the campaign upon. After you are done building, you give it away and from that point on you are told what to do with it.


Here on the other hand they let Michael build a team of teams. No one has ever built in U.S politics a team which is focused only on the concept of content testing. In Michaels’ team they had experts in email marketing, in advertising, in websites; the team was multidisciplinary, and everyone were able to contribute with very different ideas.


They also hired a person specifically to communicate, and tell the story of what they were doing in a way which would actually resonate with human beings. It is easy to get caught up in talking in a too technical way, in analytics and data all these things that people don’t really understand. So his job was to simplify it, so it can help people understand better. They created a report of what worked and what didn’t work for them during the campaign.


The information was very interesting, showing how people actually behave on their site. This got people really engaged, people shared a lot of ideas because of these storytelling reports, which they actually enjoyed reading. They also wanted to communicate to the entire organisation that any crazy idea is welcome and to encourage them to contribute. So they encouraged and tested crazy ideas. To achieve this they had to establish trust from the beginning, also with the digital director of Trumps’ campaign.


People from different organisations joint to build a campaigning operation from scratch. Employees from Brad Pascale (Giles-Pascale) who lead the Trump Digital campaign team, together with hired employees from Facebook, Twitter, Google, and their advertising partners joint by the POET team from the RNC.


Important is that they had to scale up straight away, to a national level. Implemented their strategies, they received 4 million dollars on the first day, their campaign knocked out any email campaign till this point. As soon as they started they were operating in an intense, high speed scale. They were spending 0.5 to 1.5 million dollars a day on advertising, mainly on Facebook, with a return of 1.5 on their ads, every single day, for about 5 months.


They used ad-technology which allowed them to try every single day a variety of about 20,000 variants of content and ad-copy. This in turn allowed them to choose, after only 30 minutes, the best performing ads and then bet on them with all their resources. This is basically how it worked for the entirety of the campaign. They quickly built a design team of 6-8 people who were generating content as fast as 200 pieces of content per day.


They were tracking people and their behavior according to different parameters, those which were changing between the variations of content they generated. They were testing colors of buttons, anchoring people according to their donation amounts, checked what works better, organic vs. sms vs. email vs. Facebook, and also tracking which kind of devices people were reaching the website from, such as PC, mobile, tablet.


Something very interesting they had to deal with was the fact that because they started late on the campaign, there were already other players and unofficial organisation which were cashing in on Trump merchandise, and they had to somehow differentiate themselves from those. And what we found was, that simply by putting a box on top of the donate form, which said that this is the official website of Donald Trump they could increase donations by about 10%. It was really simple things like this which were the most interesting to find out.


In terms of the internal communication it is very important not to get caught up in the technical terms, give the information in a very straightforward way in a language that actual humans understand, simple and clear. In doing that, when you are talking about website traffic, you need to be able to put a face on a website visit, as well as a dollar sign. The one key performance indicator they always got back to was revenue per visitor, practically meaning, donations divided by number of visits to the website. The point was of course to increase this number and get a better return on the money invested on advertising.


They were able to target people based on their last donation date and the amount donated and therefore, for example, they could increase the minimum donation amount they see on the page the next time they visit it.
Another thing which helped them getting our results was to run a consistent and simplified playbook. In this playbook, ads, emails, website content, and donations, are all the content that fuels the testing cycle. This cycle is a scientific method which they used to test with. starting with a question, and then based on some level of knowledge, creating a hypothesis which then can be tested further.


The 3 phases of the cycle are as followed:


  • From the offset, they were looking for the lowest hanging fruit, they were testing calls
    for action, which words to use, button colors, background images, just very simple
  • As a second phase, they were looking into developing that knowledge base, and
    what they know about their supporters, what this “brand” really looked and functioned like online. They were geo-targeting and testing state-specific messages, and then checking the results and how they can move people more efficiently through their marketing funnel.
  • And the third phase was really taking these results and learning, and optimising them across the different audiences. Then they tried to apply only to one medium of influence at a time, and check the individual results of that channel.


One thing which worked very successfully were up-sales. When people made a donation, they were offered an up-sale to increase their donation, make it a recurring donation, or at least save their donor information for the next time they visit, so they can donate then with only one click. They tried to make it simple for people to donate more, and more often.


This exposed the wizard behind the curtains. There are many deficiencies in how we measure political views and opinions. Polling is simply not accurate. If you test while people are not realising they are being tested, like online behaviour in this case, you can see their actual behaviour and decisions. This is not how it works when you are polling people, some people even said the did not support Trump when asked over the phone, while they actually did, simply as a result of social pressure, they did not want to be recognised as a Trump supporter, and that of course skewed a lot of results. The POET team on the other hand, was able to know more accurately how people behave and what they really think, get the feedback immediately, and then act upon it, and in a timely manner.


Looking back at this, taking learning from it, and applying it to the future, one realises that this technology can of course be used for good and for bad. What we have learned, is that our goal is to make this better the political process, because it can help listening to the people better and understand better what they want. As an organisation you need to know what people want in order to satisfy them. In order to know what they want you have to establish relationships, and you have to build trust so they will feel comfortable sharing with you what they are really passionate about.


Coming out of all of this, Michael opened a company, ArrowIQ, which helps people and empowers them to understand their audience better. They take their experience combining politics and technology, and using all of these tools, to help organisations, people, andparties to build relationships and positive feedback loops with their supporters, and to be able to understand them, deliver to them more personalised content, and therefore get the most out of these relationships.