Suppose the undecided vote breaks for Harris

r = Republican vote share

d = democratic vote share

u = undecided vote share

s = share of undecided vote that breaks for Harris

Putting aside the third party vote, Harris wins if

d + su > r + (1-s)u

or

r – d < su – (1-s)u = u(2s-1)

If s = .65, u = .02 (2% of voters are undecided and they break 2 to 1 for Harris) then the undecided vote would let Harris overcome what would otherwise be a loss of .006 or less in the popular vote (that is, she could overcome what would otherwise be a loss of, say, 49% to 48.4%).  If the undecided vote is .04 and it breaks 2 to 1 for Harris, she would overcome a vote share deficit of 1.2% in the decided vote (say, 49% to 47.8%).

My assessment:  could flip a close election, but turnout will matter more.