Historically third party candidacies fade into the night. What voters tell pollsters in the summer and fall is not the same as an actual vote in November.
Most of these third party flirters end up voting major party, as they want their votes to count. I would be surprised if Nader polls higher than 1% nationally. Few of any of these voters would vote for any electable Democrat. These folks are the Kool-Aid drinkers.
A look at some past Gallup polls:
2000
Nader ended up at 2.7%
1996
Ross Perot's slide took place much earlier than is typical for third party candidates. But there is a slide to be sure: He goes from a peak of 19% to less than 9% on election day.
1992
Perot's peaked at 39%, leading the presidential race in June, to 19% on election day (though the final poll was a little light for Perot). I hypothesize that because the media treated him as a major candidate, with all the scrutiny that comes with that designation, Perot's third party slide was predictably earlier than most third party candidates in the past.
1980
At one point John Anderson was polling at 24%, threatening to hit second place. He got less than 7% of the vote.
1968
George Wallace was up as high as 21%, but finished at 13.5%.
1948
Henry Wallace was polling as high as 6 or 7%, but eventually won 2.37%. Strom Thurmond was not listed in the Gallup Poll that year.