Despite Donald Trump spending 35 percent less than Mitt Romney on campaign advertising, 2016 was a banner year for political marketing.

Political ad spending hit a record at just under $10 billion despite Trump dropping $340 million less than his predecessor GOP nominee on his winning White House run. Meanwhile, the Clinton campaign spent about $1.1 billion on ads, roughly the same as President Obama did in 2012.

That’s according to a new report from Borrell Associates, which estimated that digital saw a 789-percent increase in spending from 2012.

Borrell Associates, which tracks and forecasts ad spending, said that a total of $1.415 billion was spent on online advertising by the local, state, and national campaigns this cycle.

That contrasts to what some consultants had been estimating would be a sub-billion dollar total for 2016, but it tracks with what the firm had been projecting. As recently as December, Gorden Borrell pegged the spend at $1.3 billion for digital.

“The lesson from this year is that you can win without spending a huge amount on television,” he said at an industry conference last month.

Audience Partners’ Jordan Lieberman, who had been one of the consultants questioning Borrell’s estimates, said the numbers show that the shift toward digital is accelerating. 

“This report is in line with expectations, given that it covers the healthy spending we saw in 2015 and the early primary season,” Lieberman told C&E. “However, the shift in market share toward digital is more profound than many of us expected. That whole thing about digital media addressability improving ROI might be catching on.”

Whether political digital firms can truly benefit from that shift is an open question. For instance, Borrell estimates that $2 out of every $5 spent on digital ads went to social media sites, with Facebook the biggest beneficiary.

Meanwhile, digital shops that staffed up in anticipation of a deluge of spending, such as Rocketfuel, Targeted Victory, and Quantcast, likely missed their marks with such a large portion of ad budgets going directly to Silicon Valley behemoths like Facebook, Twitter and Google.

In their report, Borrell’s analysts laid out the challenge for campaign-centered digital firms: “There are just far too many other choices – many of them accompanied by hordes of demographic and behavioral data – for today’s campaigns and causes. These choices allow political marketers to be supremely efficient with their messaging at the expense of some mass media plays.

“Political advertising is no longer just bought. It must be sold. In this new era, smart media outlets present campaigns with a thorough knowledge of the audience they provide, and the reason this audience is important.”

Overall, this cycle saw a campaign marketing increase of 4.6 percent over the 2012 presidential election with a total of $9.8 billion spent on advertising when local, state, and nationwide election spending is taken into account.

In addition to digital, another winner from 2016 was cable television, which benefited from attracting dollars from local candidates and ballot issues, according to Borrell.

In fact, cable saw a 52-percent increase over 2012, even while there’s debate raging among consultants over whether television is still the best medium to reach voters in the age of streaming services and a decline in live-viewing audiences. Cable’s share of the ad pie grew by approximately 14 percent with some $1.35 billion spent while broadcast TV saw a decline in spending of almost 20 percent.

Meanwhile, two other old school channels saw increases in spending over 2012. Telemarketing and direct mail were also winners in 2016. For direct mail, spending on postage and handling went from $285 million in 2012 to $301 million in 2016 – an increase of close to 6 percent. Meanwhile, telemarketing saw spending increase by roughly $3 million while its share of the ad market shrunk slightly from 6.4 to 6.2 percent, according to Borrell’s figures.

For 2017, the firm estimates that $5.8 billion will be spent on political advertising.