Larry J. Sabato has an analysis of the size and meaning of the “post-convention bounce” in favor of each of the two parties’ candidates going back to 1960, as measured by the difference between before and after convention Gallup polls. Sabato’s charts show a range of -1 in 1964 to +28 in 1992 for Democratic conventions. ’04 was a zero bounce year for Dems. But Sabato points out,

The size of the bounces can be deceptive in predicting the November winner and loser. Nixon’s big 1960 bounce led to a loss, while his nearly equal 1972 bounce resulted in a landslide. Similarly, Jimmy Carter’s 1976 bounce of 13 percent presaged his triumph, but his 12 percent gain in 1980 couldn’t stop a landslide defeat. Also, George Bush’s miniscule 2004 bounce of 2 percent didn’t prevent his victory.
Bounces can fade quickly. Historically, this has been truer on the Democratic side. Jimmy Carter slid from 63 percent after his convention to 51 percent on Election Day 1976, Michael Dukakis from 54 percent to 46 percent in 1988; and Bill Clinton from 59 percent to 43 percent in 1992.

In other words, despite the abundant media coverage about the post-convention bounce, it won’t mean much — “meaningless in a predictive sense,” as Sabato says.