Given how you need 15% in the first round in order to move to the second round, Sanders supporters are going to play a key role here.
Buttigieg, too. Getting 16% statewide would still mean falling below 15% in many individual caucus-sites (and getting 13% statewide would still mean being above 15% in many individual caucus-sites). What the 15% threshold mostly does is amplify the victories of the leading candidates at the expense of the other candidates. Someone polling at 5% or less statewide right before caucus day will end up with near-zero support showing up in the result because they will miss the 15% cutoff almost everywhere.
One thing to note is that Iowa will be providing raw vote totals for the first time in order to prevent some of the complaints from 2016.
In fact, when Ann Selzer was on the fivethirtyeight podcast a couple of months ago, she said that they'll release three sets of numbers:
1) The raw total of initial voter preferences
2) The total of final voter preferences after supporters of unviable candidates have reallocated themselves
3) The "state delegate equivalent" numbers that determine delegate allocation (of course, this will likely track pretty closely with #2)
Previously, only #3 was actually tabulated and released. So who is the media going to crown the "winner" if there's a split decision between the different tabulations?