It’s rare that candidates get to see a batch of television ads from their opponent’s campaign before they air, but that’s what happened in the Kansas Senate race.

Plumbing company owner Bob Hamilton is one of nearly a dozen Republicans running for the seat being vacated by retiring Republican incumbent Pat Roberts. The political newcomer has been airing television ads since the beginning of May, but an online survey with five, fully-produced unaired ads was sent out last week and one survey found its way to someone supporting another candidate in the race.

Getting feedback on ads is commonplace in big races, particularly for statewide and presidential campaigns. For decades, that was done with an in-person focus group where a moderator shows the potential ad on a screen and asks for feedback. While more ad testing was moving online even before the coronavirus changed social protocols, it’s rare that opposing campaigns and reporters get to see ads before they are on the air.

In this case, potential respondents were texted a link last week to a survey where they could watch five different ads. After each one, the person was given a multiple-choice question, “Did the advertisement make you more likely or less likely to vote for Bob Hamilton? If the ad made no difference on your vote, please just say so.”

At the end, respondents were asked to choose a favorite. “Of the 5 ads shown, which advertisement stood out to you the most? The video about making puns about what he’ll do in Washington. The video about not being establishment and fixing Washington politics. The video about illegal immigration, sancturary [sic] cities and China. The video about not being a sumo wrestler and not giving away pizza and cable. The video about law & order, riots, and the media. Not sure.”

The Hamilton campaign released two of the ads (or very similar variations) subsequent to the testing survey.

“That’s Reality,” featuring Hamilton’s views on illegal immigration and China, began airing in the Wichita and Kansas City media markets on June 7, according to tracking by Kantar/CMAG, and was uploaded to the campaign’s YouTube channel the next day. Another ad, “Law & Order,” which is similar to another of the five tested ads, began airing this week in the Topeka market. CQ Roll Call has recordings of the other three positive ads as well.

The stakes are high in the Aug. 4 GOP primary. Former Kansas Secretary of State Kris Kobach is running, and his nomination would jeopardize the seat Republicans have held for a century, considering he lost the 2018 gubernatorial race with the worst showing for a statewide Republican in at least a decade.

Rep. Roger Marshall, who represents the 1st District, has been in the race since virtually the beginning. But he was never able to put the race away with former Kansas congressman/U.S. Secretary of State Mike Pompeo’s potential candidacy consistently being discussed until the filing deadline passed last week. Former Kansas City Chiefs defensive end/Kansas Turnpike Authority Chairman Dave Lindstrom and seven others are in the race as well.

The GOP nominee will likely face Democratic state Sen. Barbara Bollier, a former Republican who outraised all of the Republicans in the race through the end of March. Inside Elections rates the race Lean Republican, but it could be more vulnerable if Kobach secures the nomination.

So how did the ads’ unintended release happen?

Since the survey would likely have only allowed the respondent to view the ad one time, the link fell into the hands of someone particularly astute and interested enough to be prepared to record the ads with their phone as they played on a laptop screen.

“Getting someone with the presence of mind to use one device to film another device in an online ad test where you only access the ad for the time it runs, and then appreciate the value of the video, then figure out how to get it to an opponent’s campaign, and actually take the time to follow through with it represents a low risk in a practical sense,” a Democratic consultant who specializes in online opinion research tools said. “But this shows it does exist.”

Similar to other public opinion polls, surveys often include initial questions that screen out members of the media or people who work in politics. But since the goal is to gauge the effectiveness of messages on voters likely to see the ads on TV, a campaign wouldn’t want to just send it to allies.

“You probably wouldn’t screen out someone just because they supported an opponent’s campaign. You actually want to get those people because you want to see if your ads persuade them,” the Democratic source said. “You’d screen out someone who works for a different campaign.”