The man responsible for spending a large chunk of Donald Trump’s digital advertising dollars isn’t a fan of programmatic advertising.
- Take our Which-50 reader survey and go into the draw for a chance to win a pair of Sony H.ear Bluetooth headphones or 1 of 5 Google Daydream VR headsets
Matt Oczkowski, head of product at Cambridge Analytica and data lead on Trump’s election campaign, oversaw a “high eight-figure ad spend” during the 2016 US presidential election.
“Not to be crazy controversial, but I have a very negative opinion of the way programmatic media is going. I think it’s a dying industry,” Oczkowski said during his presentation at ADMA Data Day in Sydney this week.
“The idea that buying a 250×250 square… you’re more likely to get bitten by a shark than click on one of those ad units.”
“I think Google and Facebook and Snapchat have it all figured out – it’s all about first party data and walled gardens. I say to any of my clients, ‘start building up your own data assets and resources and become like those guys’ – they’re never going to give you their data because you have to buy it. It’s all part of the machine.”
Oczkowski singled out native advertising, conversational Twitter advertising and Snapchat as ad formats that worked well for the Trump campaign.
After joining the Trump campaign in May 2016, Oczkowski and his team built up the Trump campaign database, conducted polling and digital ad buying from an office in San Antonio, Texas.
“His campaign was almost entirely data-driven — outside of Mr Trump because he does his own thing — but everything else was very data-driven,” Oczkowski said.
Specifically Oczkowski was focused on just 15 million people within the overall US voting population of 230 million: the people they could persuade who lived in battleground states.
Oczkowski emphasised the need to have your own first party data sources, “There’s no market difference when you are all working off the same data sources,” he said.
Based on donations, sign ups, store transactions, volunteers, event attendees, access to first party data was a major advantage in its targeting, he said.
Trump’s data team built models to solicit donations, persuade undecided voters and encourage people to get out and vote.
Data also heavily influenced Trump’s travel schedule in the last two months, Oczkowski said. His team built a priority cities tool that ranked all the cities in the 17 battleground states Trump was trying to win.
The dashboard was updated when they received new polling data and the rankings of cities would change based upon the importance of winning voters there. The calculations were based on the mix of persuadable voters and core supporters, “because you don’t want to go to a city with a bunch of ambivalent people that aren’t willing to cheer.”
Oczkowski also critisicised the Clinton campaign for following the Obama 2012 campaign playbook too closely.
“The Obama team in 2008 and 2012 really revolutionised technology, politics and government,” he said.
“But a lot of the practice today comes from Obama’s playbook and I think Clinton and her team followed this playbook almost to a tea which I think ended up being a bit of a detriment this year because of the election was just so different.”
Instead Oczkowski advised campaigners heed the advice of Obama campaign CTO Harper Reed and “blow up your playbook soon as the campaign finishes” not try to copy it four years later.