Great point! That's the question I'm asking myself too.

I'm erring on the side of moving on too quickly, rather than waiting too long, because I like to move fast and I've already made the mistake of spending way too long on something.

I think it's really non-obvious, and there aren't a lot of good heuristics that measure how long is long enough -- and it's my belief that most heuristics simply aren't aggressive enough. One of my mottos is "most limitations are conceptual and self-imposed," and I see a lot of self-imposed limitations in the SaaS community, mostly because people have no idea what can be achieved when you set really high goals... not sure why.

What is your mental model for how customers exist? Because 2 years to pivot and manage to validate a SaaS business seems like a reasonable rule of thumb here, though naturally working 2 hours/week can cause people to take longer. Working 100-hour weeks for 3 weeks may shorten the timeline to go from idea to (working!) MVP, but it doesn't change anything about other people.

Consider trying to sell a car or other large purchase. The equivalent to 90-hour weeks on the software would be spending however much time detailing/cleaning the car. There are some fundamentals about customers that detailing the car won't affect. Some people aren't, and will ~never be in the market to buy a car. Some people aren't in the market to buy a car right now. Some people don't, and aren't going to want that kind of car (no matter what it is). Some people don't (and aren't) going to have the money to spend on a car (or that kind of car), no matter what. (But they'll still come around to kick the tires.)

In order to sell the car, you have to get lucky, but even then, there's a semi-fixed amount of time between when someone decides to buy a car, to when they actually buy a car. If they manage to see your car in that period, then you're in luck, and can make the sale, but there's nothing to do that can force a customer to be in the market to actually be in the market to buy a car. (In selling a car, there are tactics that used car salesmen will employ to try and get a customer to buy a car, today, but working 90-hrs/week isn't using those tactics in this metaphor.)

Ultimately, my point is that there's a lot of hurry-up-and-wait in business, and 3 weeks isn't long enough of a wait. Working 100-hour weeks shortens your timeline to get to a product, but it doesn't shorten your customers'.

Customers need to be exposed to the idea of your product, and exposed to your product, and be repeatedly exposed to its existence, a number of times, over several months, in order to be curious enough to try the product. From those that have tried the product, the percentage of those potential customers, for whom the pricing model works for, is slimmer still.

So then, what should the cutoff be? Whatever that cut-off should be needs to be backed by some thought as to how long it will take a customer to manage to engage with the product. That model must take care to have the whole picture. Validation for a bar, who's product practically sells itself on a Friday night (pre-pandemic), isn't timing how long it takes a patron to come in the door and then order a drink, but is a much bigger question of validating that the bar's culture fits in with the location that it's in, and involves being open for O(2 years) in order to see what "normal" sales are like for a given holiday.