Thanks 🙂
Are you running this fallout on "this month" data (or really anything that includes today)? That could account for the top level variance of about 11 people (if you have data actively coming in, the time at which each visualization was calculated, there could be differences).
However, I think that the reason your "Site Section equals...." step is so far off is because there is a difference in the pathing...
On the "segment as step" version, the user must hit the segment first, then they hit the section (and that section doesn't necessarily meet the hit criteria of your segment...)
Example: HIT segment page title contains "abc"
Page A > Page ABC > Section Page ABC
OR
Page A > Page ABC > Section Page C
would both match, the segment as a step catches the second page, and the Segment doesn't necessarily have to contain ABC; but it must hit two specific pages in sequence... ABC then a section page
On the other hand, in the segment applied to the whole visualization, all steps inside the sequence have to have a page title that contains "ABC" (which you would think would make the numbers lower), but there is really only a single step here... there is no need to hit 2 pages, one is enough to meet the criteria.
Page A > Page ABC > Section Page ABC
OR
Page A > Section Page ABC
This doesn't require hitting Page ABC prior to section ABC.... you can even see that your second fallout is missing "EVENTUAL PATH". So it's would be lower due to the higher restrictions of the Page Title on the section, but higher since all direct traffic to those pages will be included.