So what are those expectations?
From our individual-level partisan models, we know that the early vote is more heavily Republican in this round of voting. By how much? 6.5%. The average modeled party score for 1st round early voters was 47.9%. In this 2nd round, the average party score dropped to 41.4%.
So what does this mean?
If Ossoff runs 6.5% behind his round 1 early vote performance (ie, wins 57.2% of the vote) he is showing no change from his round 1 performance which yielded him 48.1% of all votes cast
If Ossoff runs further ahead of the generic party model, exceeding 57.2% of the early vote, this is an early sign that things are going his way
If Ossoff runs further behind modeled partisanship, failing to reach the 57.2% benchmark among all CD6 early voters, this is a sign that he's in for a steeper uphill climb among Election Day voters
Simplified — When the early vote results are tabulated, think of 57.2% as the target for Ossoff.
Let's break that goal down by county, using the same method (normalizing Ossoff's round 1 performance by the change in modeled party turnout in round 2):
