They do better the further away from the election you happen to be. Two variables drive that generally predictable decline: (1) Many of the early undecideds pick "the other guy" rather than to say that they don't have a preference; and (2) the two-party candidates have not yet spent hundreds of millions in advertising to convince every last person that their third-party vote could make them a spoiler.
Nonetheless, if you measure the yardstick of success correctly, even early polling data is important, and useful.
I would count the Gary Johnson/Jim Gray ticket as the most successful in Libertarian Party history if it achieved 1,000,000 votes in the General Election, which would amount to roughly 8/10s of 1 percent of the total vote in 2008.
I would count the ticket as wildly successful if it achieved 1 percent of the vote in the General Election, which will probably be somewhere around 1,400,000 votes.
I would count the ticket as fantastically successful if the Libertarian candidates were actually invited into the contrived joint campaign appearances wrongly called "presidential debates."
I would count the ticket as successful beyond my wildest dreams if it garnered the guess-timated 6,500,000-7,000,000 votes in the General Election that would probably be necessary to reach 5% of the popular vote and thereby qualify the LP for tens of millions in Federal matching funds in 2016.
(I would be willing to live through four years, I think, of either Obama or Romney having won because Gary Johnson was a spoiler, because $90-100 million in Federal matching funds would be such a game-changer in American politics. I realize that in this opinion I part company with a great many Libertarians, not to mention Republicans and Democrats. But I would add an addendum that, in the even the LP did qualify for Federal matching funds in 2016, Wayne Allyn Root should be the subject of a special rendition and be placed in the Bagram, Afghanistan prison for that entire year.)
So those are my standards.
Here is the early polling data, such as it is.
On 19 April, Public Policy Polling found that in a three-way race, President Obama would get 47%, Governor Romney would get 42%, Governor Johnson would get 6%, and Undecided would account for 5%. Of course, the same poll determined that only 25% of the respondents knew anything about him, and among those there was a 3-1 negative opinion. So one could take the optimistic approach: Johnson has legitimate spoiler cred, or one could take the pessimisstic view: of the people who hear about him only 25% will even consider voting for him.
On 25 April, PPP published a poll on Johnson's home state, New Mexico, which was reported by Ballot Access News as finding the following:
The results: Obama 48%, Romney 35%, Johnson 15%, other or undecided 2%.
The poll shows that 10% of Democrats, 20% of Republicans, and 23% of independents and members of other parties choose Johnson.
One would assume Johnson would do better in his home state than nationwide, but 15% probably stretches the far limits of election day credibility. What the poll does begin to shed some light on, however imperfectly, is that idea that Johnson (a former Republican) could have as much appeal to Independents as Republicans, and could draw a smaller but still significant percentage of Democratic voters. Being able to draw Democratic voters is critical to Johnson's credibility as a real candidate and not simply a Ralph Nader-Pat Buchanan spoiler.
The New Mexico poll is also interesting because there 79% of the respondents report themselves as familiar with Johnson, but he still loses the favorability ration by 42-37%. This, however, seems to correspond to the fact that most New Mexicans know Johnson as a Republican, and his numbers in that regard are not dissimilar to Governor Romney's in a state that President Obama is currently seen as winning.
There is very little else out there yet with regard to Johnson's numbers, but these early ones are sufficient to raise an important question of political strategy. The answer will go a long way toward determining if the Johnson/Gray ticket meets any of my standards of success for the LP candidate in 2012:
Will the Johnson campaign team be able to conduct the sort of analysis necessary to give the candidate data--not just opinions--regarding which voters are more likely to support him, and why?
Seeking 1,000,000 votes in a potential General Election voter total that could exceed 140,000,000 is a completely different proposition than Obama or Romney face, even leaving aside the huge disparity in resources. Romney has to move from the retail politics of the primaries to the wholesale politics of the General Election, attempting to tailor his messsage to be congenial to the widest possible audience. Governor Johnson is never going to be in the position to conduct even retail politics--he is in the boutique category. He has to identify specific concentrations of voters who have the characteristics most likely to make them receptive to his message, and then he has to go out and get them. Personally.
Randomly traveling around the country making small-scale campaign appearances among the faithful will not cut it.
To get 1,000,000 votes or better, Governor Johnson's campaign needs something that other Libertarian presidential campaigns have generally disdained: a data-based political strategy.