For decades, White House pollsters for Republican presidents have been paid by the RNC, while the Democratic National Committee has paid for presidential polling during Democratic administrations.
But the prolific output of polling data from America First Policies has coincided with a steep drop in the amount of money the RNC spends on presidential pollsters, as compared with the first 12 months of previous Republican administrations dating back to the Reagan White House.
According to FEC reports, the total amount that the RNC spent on Trump's campaign pollsters in 2017 amounts to just 5 percent of what the committee spent on Ronald Reagan's White House polling in 1981, adjusted for inflation, and around 12 percent of what it spent for President George W. Bush's pollsters in 2001.
This is partly due to a shift in the type of data system that is used by the RNC, away from traditional polling and toward a scoring system, according to an RNC official who spoke to CNBC on the condition of anonymity.
"Since 2013, we've spent $250 million to gather information through voter scoring, and we have a huge amount of information that informs these scores," the RNC official said. "So, we don't really pay for traditional polling anymore. We rely on this data, instead."
On the rare occasions that traditional White House polling is needed, the RNC official said, the committee still relies on the president's preferred polling firms. This happened twice in 2017 and once again in January.
Trump's top polling firms, meanwhile, do not come cheap. The three firms currently identified on some of America First Policies' documents charged the Trump campaign a combined $2.1 million for just two months' worth of polling during the final stages of Trump's 2016 presidential campaign, according to FEC filings. At these rates, the cost of the polling these firms have produced for America First Policies since last summer could easily be in the seven figures.
Yet ever since Trump was inaugurated, these same filings reveal that the rebooted Trump 2020 campaign has not spent any money on polling, despite raising $17 million last year.
"This looks like a million-dollar polling operation," said Spaulding, the former FEC special counsel, who reviewed the documents that CNBC downloaded. "And this isn't the kind of issue polling you'd expect to see from an outside nonprofit: This is more like a shadow campaign on behalf of the president."
Much of the polling itself is deeply flawed, however, according to polling experts who reviewed the documents uncovered by CNBC. For example, America First Policies' pollsters frequently use leading questions, which make it clear that one answer is much better than other possible answers. The group also employs leading questions to test public opinion, skewing their own results and painting an incorrect picture of how voters feel about certain policy issues, like immigration and infrastructure.
"If you want to test a position, you test a position. If you want to test a message, you test a message. There's a time for both. But you don't deliberately f--- up the results of your position survey by dumping a message in front of it," said a veteran political pollster who spoke on the condition of anonymity. "If you're trying to gauge a person's stance on asylum seekers, you don't taint your question with some bull---- about how violent MS-13 can be before you ask it. I mean, what good does that do? All it shows you is that scary things can scare people."
America First Policies' Montgomery acknowledged that some of the questions were leading, but said the ones she had helped to write were not leading. Walsh said the leading questions were actually just message testing. But message testing typically occurs only after public opinion has already been measured without the use of leading messages. Many of America First Policies' questions contain leading language the first time they appear on a survey.
One recent example of leading language was in a February poll's description of Trump's proposed military parade. Here, voters were asked whether they supported or opposed "a military parade that honors our nation's veterans and the troops and families who sacrificed their lives for our country, as well as the strength of our military."