Sounds accurate. After SurveyUSA cleaned up their act in Kentucky in 2015, I tend to trust them more than I used to.
They said Conway was going to win. And they were way off downballot as well.
Other polls (like WKU) agreed with SurveyUSA.
It wasn't the polls that were off. It was the election that was off.
The polls should have been better at modeling turnout then. Claiming that the polls were fine because they "modeled what was supposed to happen" is at best a misunderstanding of what polls are to be - reflections of what WILL happen, not reflections of what would happen in an unusual, freak situation (since Bevin won by so much , him winning is not a freak situation), and at worst a pure partisan hack viewpoint.
SUSA is a company is not bad, but they are simply unable to model the KY electorate, and that's why Bluegrass banned SUSA from doing any more polling for them after the 2015 elections.