What is an example of polls being an inaccurate in determining voting behaviour?

Most polls in 2017 predicted a comfortable conservative majority a few days before the election, perhaps the polls ignored the youthquake from young people turning up to vote Labour