One thing we found early on selling to universities was that if we priced ourselves to low, some schools wouldn't give us the time of day, other schools would start comparing us to "competitors" who weren't our competitors at all and did nothing like we did, but who shared a similarly low price.
I still remember when we sold our first (small 3000 student) school for $6000/yr. When they gave us the check they said, "Eh, that's a drop in the bucket." Talk about leaving money on the table.
If enough people do drop the original one and make the premium package the 'normal' one, offer a discount to your old users.
People aren't going to spend more for the hell of it. I don't think this is a useful technique for determining the minimum viable cost for a service.
Even the same model car but a more expensive trim will sell for a higher price than the same model without the trim.
Features can be used to substantially increase the price of a product while only marginally increasing the cost.
So, is there a way out?