It works fine over in the cell phone market, where carriers market themselves in terms of HSPA/4G/LTE/5G while plans are sold in terms of data used. Rough instantaneous bandwidth (and latency!) is fine for knowing what applications will work. 300Mb/s vs 500Mb/s isn't going to make much of a difference for most, and certainly not more than ISPs necessarily massage the numbers when talking about shared mediums like docsis/gpon.
The point I am making is that if fixed carriers seemingly no longer want to be selling connections by last-mile bandwidth (with implied statistical oversubscription and a fall-back to best effort), and instead want to move to selling specific amounts of data delivered, then their big-print marketing should have to reflect that!
Or alternatively, if they want to stick with selling the instantaneous last mile connection speed, then data quotas should have to raise along with that according to a standard benchmark consumer usage - say 2 hours of nameplate usage per day, meaning a 300Mb/s connection would be ~6TB/month. And then additional data usage should cost the same (or lower) as the monthly plan - for example $120/month for that 6TB/month plan is $20/TB, so 1TB of additional data should cost no more than $20 rather than the current punitive or even unobtainable pricing.
Also I'd guess your 90%/1TB figure is quite out of date, which gets to the heart of the problem here - these limits have remained arbitrary and opaque rather than being expressed in the market. It gets even more rotten when talking about national policy where these ISPs have been able to claim they're providing modern "broadband" while their backhaul networks have been seemingly frozen in time for a decade+.
Although, a disclaimer: I only read these threads to gawk at the train wreck and check if the Internet access market is really still this bad. I've got municipal gpon at 1Gb/s symmetric, no caps (I've asked multiple times), with a fixed price every month. Which really just underscores the absurdity of talking about this topic in a negative-sum parceling out of bits and bytes, which mostly stems from Ma Coax wanting to continue milking their captive markets on aging infrastructure.