Everyone was wrong about 2016. Then they were wrong about being wrong.

Today, more than a month past Election Day, the 2016 election officially ends; the states have certified the results, and the Electoral College will almost certainly vote to install Donald Trump as the 45th president of the United States.

Most — including experts, pollsters, and even some in Trump’s own campaign — didn’t see this coming. When it happened, there was a scramble to figure out how it did.

In the days following Election Day the what happened in 2016 story was largely about who didn’t come to the polls; Clinton was simply so unpopular that she couldn’t mobilize people to come out and support her, reports said. Trump had won with record low turnout.

“What you can’t do is you can’t manufacture enthusiasm,” David Axelrod told the Washington Post about a lack of energy in Clinton’s base. “There was an assumption that antipathy toward Trump would be enough to mobilize the base ... a certain lethargy that sets in when you’ve had the White House for eight years. Your troops are just not as hungry.”

But then the story shifted. Voter turnout in 2016 was actually closer to 58.9 percent, slightly higher than 2012, according to data from the US Elections Project. Clinton won the popular vote by more than 2.8 million votes — more than Al Gore won the popular vote in 2000, John F. Kennedy in 1960, and Richard Nixon in 1968. She lost the election by 107,000 votes in three states.

So how did everyone get it so wrong — and then get it wrong about getting it wrong? The simple answer: It takes a long time for all the votes to be counted.

On November 9, the election night takes were in. Clinton’s weaknesses had come full circle: She was an unpopular, too corrupt insider, too uninspiring to win, even in the face of an inexperienced bombast running a divisive campaign on exclusionary policies.

On November 9, the sentiment was that Republicans had won with a mandate; they took the White House, the House and Senate, and governorships in 33 states.

Then, day by day, the mandate weakened. There were millions of ballots uncounted. On December 1, California still had 4 million ballots left to count, many of which went to the Democratic candidate. Clinton’s popular vote win grew by the millions. It turned out there wasn’t a huge fall in voter turnout from 2012, and it wasn’t the lowest turnout in 20 years.

Democrats were still able to turn over two Senate seats. And despite all the Trump hype, interest in third parties should not be understated this election.

In 2012, Gary Johnson attracted 1 percent of the national vote. This year he pulled just over 3 percent. Jill Stein got 1 percent of the vote this year, compared with less than half a point in 2012.

That translates to a lot of ballots in battleground states, according to Vox’s analysis. Florida saw an increase of 185,000 votes for minor parties between 2012 and 2016. In Wisconsin, there was an increase of 105,202 votes this year, relative to the last election cycle. Basically, minor party candidates tripled their support this year — a testament to the unpopularity of both major party candidates.

Clinton’s popular vote win made Democrats feel like she should have won the election; at the very least, it provided a slight comfort in an otherwise devastating loss.

It also proved that Clinton, the “unpopular” candidate, was able to turn out voters — just not where it mattered.

Democrats lost the presidential election in three states; Clinton lost in Wisconsin by fewer than 23,000 votes and in Michigan by fewer than 11,000 votes. Votes afforded to Green Party candidate Jill Stein in either state were more than Trump’s margin of victory. Trump won Pennsylvania by a slightly bigger margin, surpassing Clinton by a bit more than 44,000 votes.

Turnout mattered in these states, according to Stanford University political scientist David Brady’s analysis, as reported by the BBC:

Voter data in Pennsylvania, Michigan and Wisconsin as well as Minnesota, found that districts in which Mrs Clinton won by more than 70% showed that though the population had increased, turnout was down.

Conversely, the districts that Mr. Trump carried in those states by more than 70% showed that population had declined, but turnout had increased, signaling Mrs. Clinton was unable to energise voters in those key states the way her Democratic predecessor did in 2008 and 2012.

Clinton’s ability to mobilize voters was in already-blue states — states like California and New York.

Her popular vote weakened the Trump mandate, but it still lost her the election.

Correction: The original article incorrectly stated Trump had a roughly 70,000 vote lead over Clinton in Pennsylvania. He in fact had just more than a 44,000 vote lead in the state. We regret the error.