A lot of stretch goals I have seen are things like upgraded components, extra colors, and the like which are wuite possible as your order sizes push your per-unit costs down. This is doubly so for software based offerings since marginal costs are zero for each additional pledge so you can take what would be pure profit to fund additional features.
...many projects don't seem to have any idea how scale works though.
All I'm advocating is that creating a project requires you to submit a budget detailing how you'll spend the funds.
Spreadsheets are easy to parse; it'd be trivial for them to automate checking the sheet for stupid values (cost of manufacture: 0, cost of shipping: 0, profit: 100%) or people who have no idea (100% of funding -> making the game (that's not a budget doofus)) or people trying to exploit the system (excessive total units shipped) and give them some idea of what projects need closer scrutiny before being approved.