On Monday night, Donald Trump and Hillary Clinton met for their first debate. Most observers, including conservative commentators, credited Clinton with a strong victory. The Atlantic Monthly called it “a commanding performance.” The New York Times claimed she scored points on issues of “race, gender and national security.” But did it matter? Or did it just cause viewers to dig in their heels?

The scholarly consensus is that debates don’t usually matter. They might shift public opinion by a few percentage points, but, by the time debate season rolls around, most voters have made up their minds. Voters watch debates to confirm their choice and to root for their preferred candidate. Even a major gaffe or a stinging one-liner is unlikely to convince those voters to shift from one candidate to the other. And in today’s polarized environment, many of those minds have been made up for some time. Robert Erikson and Christopher Wlezien examined polls conducted between 1952 and 2008. With the exception of the 1976 debate between Carter and Ford, polls before and after the debates show “a fairly strong degree of continuity.” In 2012, Barack Obama was seen as having won his first debate with Mitt Romney, giving him a boost in the polls, but, as noted by John Sides and Lynn Vavreck, the effect was fleeting, due in part to Obama’s weaker performance in the subsequent debate.

So does that mean that Clinton’s performance doesn’t matter? Not so fast.

Recent weeks have seen a tightening of the presidential race, with many national and swing-state polls showing a race that appears to be neck-and-neck. While strongly affiliated Democrats and Republicans are unlikely to vote for the opposing party’s candidate, many swing voters with weaker partisan ties are still making up their minds.

For undecided voters—those less likely to have a strong partisan preference that predicts their voting choice—the debates may be the first time they are tuning in to the race. And, Monday night, those voters not only tuned in, breaking a 60-year old viewership record for presidential debates, but they also stayed tuned in, watching not only the first 20 minutes (when Trump was still doing well), but the whole thing, contrary to expectations. Despite pre-debate concern in some quarters that the debate was competing with a Monday Night Football game, the debate drew ten times the audience.

In other words, many undecided voters were watching when Trump was roundly defeated by Clinton—and they were likely influenced by what they saw.

Google searches in Spanish for “register to vote” (“registrarse para votar”) peaked during the debate, suggesting that many Latino viewers were inspired, at that moment, to get involved. Given the state of pre-existing Latino preference for Clinton, those voters are also likely swinging toward the Democratic nominee.

Do unregistered Latino voters matter? As Fernand Amandi wrote recently in Democracy, the answer is: Absolutely, they do. In swing states, and even in some traditionally Republican states like Arizona, higher rates of Latino registration and turnout could turn a GOP-leaning state or a tied battleground into a win for Clinton.

Younger voters with weaker party preferences than older voters of the same ethnic group have also tended toward the undecided column this year. They may have been rooting for another primary season candidate, such as Bernie Sanders, or they may have been leaning toward a third-party option like Libertarian nominee Gary Johnson or Green Party nominee Jill Stein. Clinton has been reaching out to young voters: for example, her decision to do an interview last week with Zach Galifianakis on “Between Two Ferns.” Harvard University’s Institute of Politics hosted a national online focus group with undecided young voters on Monday evening. At the end of the event, twice as many said that they now planned to vote for Clinton compared to Trump.

Overall, of those voters who have said that they plan to vote for someone other than Clinton or Trump, many were likely also influenced by the absence of third-party candidates from the debate stage. In 1992, Ross Perot was polling strongly enough to join George H.W. Bush and Bill Clinton in the national spotlight; without that general-audience visibility, voters considering a third-party vote now likely see such a choice as less viable.

Another group influenced by Monday night’s head-to-head were suburban women, a group Trump has struggled to connect with. And he certainly didn’t make much headway at the debate. Defending his characterization of actress Rosie O’Donnell as a “big, fat pig” a decade ago, Trump said, “she deserves it, and nobody feels sorry for her.” Clinton also attacked Trump for his comments about former Miss Universe Alicia Machado. On Tuesday, Trump doubled down on those comments. “She was the worst we ever had,” he said. “She gained a massive amount of weight, and it was a real problem. And not only that, her attitude.”

Trump’s numbers have been lagging among women voters, and these comments did him no favors. Undecided women are now even more likely to vote for Clinton. They are seeing, in Trump’s treatment of Clinton and of other women, the sexism that they have experienced in their own lives. Clinton didn’t need to call it out; the “woman card” was played for her.

Finally, more than in any other campaign in recent history, voters who do show a clear preference between Clinton and Trump have said they are voting out of a strong desire to stop the other candidate from becoming President. And this week’s debate likely only reinforced their preference. Many potential voters—unregistered voters as well as registered but undecided ones—likely saw something on Monday night that may have nudged them to get involved or to finally make a choice.

Yes, Clinton won. Decisively. Will it matter? Almost definitely. Unregistered voters, especially Latinos, are exploring the voter registration process. Women are taking note of Trump’s sexism. Johnson and Stein fans are taking a hard look at how such a protest vote might matter in a tight election.

As already mentioned, the scholarly consensus says that debates mostly don’t matter (with an asterisk for 1976). But this year’s presidential election has already broken a number of assumed political science “rules.” Given the content of Monday night’s debate, that consensus might also need to be revisited.