In the final weeks of the Minneapolis mayoral campaign, Mayor Jacob Frey's top challengers had a unified plea to voters: Leave Frey off their ballots. But a Star Tribune analysis of new data shows the most common strategy voters used instead was to pick Frey — and only Frey.

Nearly 27,000 voters — about 19% of the total — chose Frey and no one else for mayor. It was the exact opposite of the #dontrankfrey campaign, which hoped to convince voters to use ranked-choice voting strategies to oust the incumbent.

"We did not tell anyone to only rank the mayor," said Joe Radinovich, Frey's campaign manager. Radinovich said they focused on a "positive message" about boosting police accountability, housing and economic inclusion.

"The #dontrankfrey strategy, we believe, was limited as a political message," Radinovich said. "That message I don't think told people what those candidates were going to do when they were elected and instead was just a message of opposition."

To learn more about voters' preferences in the November election, the Star Tribune analyzed anonymized voting data that shows the individual combinations marked on voters' ballots. That information was released late last week, after the city certified the election results.

Two of #dontrankfrey's most prominent supporters were mayoral challengers Sheila Nezhad and Kate Knuth, who banded together urging voters to rank them in the first and second slots. They argued it was time for a new leader who would do more to improve public safety.

While the #dontrankfrey campaign lost, it resonated in parts of the city — and particularly in areas where there was strong support for the ultimately unsuccessful proposal to replace the Minneapolis Police Department.

The second and third most popular vote combinations placed Nezhad and Knuth in the first two slots and left the third ranking blank. Nearly 20,000 voters went that route.

The strategy didn't benefit the candidates evenly. Despite earning more than 30,000 first-choice votes, Nezhad was eliminated in the first round of ranked-choice voting tabulation. Knuth made it to the final round.

To understand why Knuth advanced further than Nezhad, it's essential to know how votes are counted in ranked-choice elections. If no candidate receives more than 50% of the vote based on first choice rankings — as was the case here — election officials use ranked-choice voting tabulation to determine the winner.

First, they looked to see who had the highest total number of first choice votes. In this case, it was Frey with 61,620. Then, they looked to see whether it was mathematically possible for other candidates to surpass him. Candidates who didn't have a chance to surpass him, or to win more than 50% of the vote, were eliminated.

There are multiple ways the city runs those calculations. The simplest involves looking at how many first-, second- and third-choice votes other candidates received.

Nezhad received just shy of 60,000 total votes across all rankings, not enough to pull ahead of Frey. She was eliminated. Knuth had just over 69,000 total votes across all rankings, so she stayed in the race.

Elections workers then looked at ballots submitted by people who had ranked eliminated candidates first. If they ranked either Frey or Knuth second, those votes got added to the total. If they didn't, elections workers looked to see if they had ranked Frey or Knuth third and, if so, added those votes to the total.

Knuth logged the highest number of second-choice votes in all but seven precincts in Minneapolis.

In campaigning, Knuth's supporters often asked voters if they would be willing to give her their first choice ranking. If not, they asked for second or third.

"The strong second choice support across the city shows how Kate connected with voters across Minneapolis," said Faisa Ahmed, Knuth's campaign manager.

While people's second- and third-choice votes proved crucial, so did the options they declined to fill out. Of about 144,000 ballots cast in the mayoral race, about half weren't filled out with all three choices. More than 70,000 voters left one or two choices blank.

Correction: Earlier versions of this story misstated the number of voters who left part of their ballots blank.