From the very first day of the very first voting in the 2016 primary process, it was apparent that Donald Trump’s political success would depend far more on his ability to rile up supporters than on traditional political tools. You’ll remember that he was favored to win Iowa in February 2016, with the Selzer & Co. poll released shortly before the caucusing showed him up by five percentage points. He ended up losing by three — far more a reflection of Sen. Ted Cruz’s superior ground operation to turn out voters than any glitch in the polling itself. In the context of Trump’s campaign abilities, it can’t be repeated often enough: Trump is the first president in the modern era to win with less than half of the vote in both his party’s primary and the general election.

This is an important gap in Trump’s résumé for the simple reason that he has insisted that he will oust elected officials who defy him. He told Rep. Mark Meadows (R-N.C.) that he was “gonna come after” him if Meadows continued to balk at Trump’s health-care replacement. Perhaps recognizing that Trump’s threat was hollow, Meadows’s opposition, paired with that of other conservatives, tanked Trump’s legislation.

Since that disaster for the administration, Trump has had two chances to flex his campaign muscles: a special House election in Kansas’s 4th Congressional District and one on Tuesday in Georgia’s 6th. Republican Ron Estes won the race in Kansas last week, replacing Mike Pompeo, whom Trump chose to run the CIA. In Georgia, Republicans are hoping simply to force a runoff race against Democrat Jon Ossoff, who has consolidated support from his party in a way that none of the numerous Republicans he faces has been able to do. That runoff, though, is likely to mean a contest between Ossoff and one of those Republicans — much more favorable turf for the party.

So what has Trump done to weigh in against Ossoff? Two of his favorite things: He has tweeted, and he has recorded his own voice.

Over the past few days, Trump has tweeted five times about Ossoff’s race. In the first tweet, he criticized the race for becoming a media event. Then, he disparaged the “super Liberal Democrat” and declared that a runoff election “will be a win.” Ossoff, he said Tuesday morning, would be “a disaster in Congress” who would “raise your taxes” — by somehow overpowering the Republican majority.

Those tweets have been retweeted about 51,000 times, combined. Twitter has proven to be an effective way for Trump to dominate media attention when he makes surprising pronouncements such as (falsely) accusing his predecessor of wiretapping his home. But these more anodyne political tweets don’t gin up the same level of interest and, given that only 21 percent of the adult population in the United States uses Twitter — skewing heavily toward younger Americans who are also less likely to turn out to vote — the odds that these tweets are motivating much of a response seems low.

Trump also recorded a robo-call in the race — one of those prerecorded messages that gets pushed to your answering machine. In it, Trump encourages Republicans to vote, but without identifying a candidate. Instead, he exhorts voters to “stop the super liberal Democrats” and keep Ossoff from “flood[ing] our country with illegal immigrants.”

In a race in which you’d expect the focus to be on Ossoff running against Trump, instead, it’s Trump running against Ossoff.

Will it work? Robo-calls are cheap and easy, which is a key reason they’re popular in political campaigns. But there’s little evidence that they’re terribly effective. In the book “Get Out the Vote,” researchers Donald Green and Alan Gerber detail the utility of various voter-turnout tools. They say that robo-calls “might help you to stretch your resources in ways that allow you to contact the maximum number of people, but don’t expect to move them very much, if at all.” Multiple studies cited by Green and Gerber showed statistically insignificant effects on turnout — even when celebrities or the leader of the party were the ones leaving the message.

What might Trump have done instead? The Georgia election is a weird one, with no anointed candidate for Trump to stand next to. What’s more, he won the district by only 1.5 points, according to Daily Kos — meaning that his popular appeal is limited. He certainly could have done an interview with a local television news station or approved a targeted piece of mail to conservative Republicans in the district. White House press secretary Sean Spicer could have invited a local reporter to ask a question remotely during a news briefing. There were ways for Trump to get involved without setting foot in Georgia that he appears not to have done, relying instead on his tweets and his voice.

Not that he did anything more than that in Kansas either. Estes’s victory was a welcome one for Republicans, but it was far closer than might have been expected several months ago. Estes won the district by about seven points — but Trump won by 27. That was a place where Trump could have helped bolster the Republican more directly, by showing up on the ground for Estes. He didn’t. The weekend before the election, instead of heading to Kansas and holding a get-out-the-vote rally for his party’s nominee, he was at his Mar-a-Lago resort in Florida — and his nearby golf course. All Trump did to persuade voters in Estes’s district was to tweet — once — and record another robo-call.

Estes’s victory was hardly the sort of thing for which Trump can take credit, given how little he did and how much worse Estes performed than he did. There’s little evidence that Trump knows how to sell anyone politically except for himself — a challenge that his predecessor also faced. President Barack Obama repeatedly tried to leverage his own popularity on behalf of other candidates, with the net effect of losing the House in 2010 and suffering humiliating losses in 2014. Oh, and, of course, seeing his preferred successor lose in stunning fashion in November.

The difference between Obama and Trump in this regard seems simply to be that Obama was better at selling his candidacy than Trump was at selling Trump’s. In many places, perhaps including that district in Georgia, Trump doesn’t have popularity to leverage anyway. Meaning that if you’re a candidate wondering if you should rely on Trump to push you over the finish line — or if you’re wondering if he’ll stand in your way — you’d be forgiven for assuming that the president and his wan political outreach probably would be a non-factor.