There are misleading points of convergence with Sagarin ratings and poll rankings, but they're really apples and oranges.
Ratings provide you with a mathematical tool for predicting outcome, at least to the point of predicting the probable point spread based on past results not only of two specific opponents, but a field of subjects producing data under the same terms.
Rankings are at based on win loss record and badly skewed by opinion. There is little to no prediction value in rankings--at least when it comes to actually deciding point spread.
The only thing Sagarin "ranks" is strength of schedule and that ranking is only a comparison of how a particular team's schedule compares to the ratings teams in the top 10 and top 30. Even that is based on an average of ratings for your opponents, not a poll-style ranking.
Anyone who suggests that Sagarin ratings are actually rankings of teams 1 through 252 misses the purpose of the scheme entirely.