The polls were pretty accurate in 2016. They had Clinton with a small (~3%) lead nationally and she won the popular vote by ~3%. Where they missed, in the Midwestern battleground states, they missed by a small amount, but the electoral college swung on a tiny number of votes. What you shouldn't put much stock in is talking head "analysts." They kept claiming that Clinton was going to win in a landslide even though that's not what the polls said.