I think there are some mis-perceptions about the idea. Here is how I how found the concept useful, in my company.

Startups who focus on revenue rather than growth can use this technique as a success metric (duh). Rather than using active users, etc., as proxy metrics for success, this method is a very direct one to measure how marketable your idea is. Get one success banked, then you can start to optimize -- pretty neat!

However, it's also incredibly difficult to do. So the second purpose becomes this: a proxy to evaluate the importance (in your customers' eyes) of the problem you solve. If you're solving a problem that everyone in the banking industry desperately needs solved, they will be more willing to try something that isn't feature-complete (or in existence).

So while few people will actually buy your product before it exists, using that as the goal can help you in your business development. You can figure out the problem's importance (and, believe me, it's better to solve a hair on fire problem than to solve a minor inconvenience). You can get feedback from people who are actually willing to pay -- this signals they are probably your customer! (So it helps you prioritize feedback, and sort people you talk to into actual / window-shopper type customers).

What the gurus don't tell you is this is very difficult to actually accomplish, at first. But after several iterations, you can get valuable feedback from people who are closer to having skin in the game.

And, of course, there is the chance that someone will have such a big problem that they want to pay you to solve it right now. If you can find that big fish, and their problem is similar enough (or uses core technology sufficiently similar to your "productized" version), then you can partially pay to make the product. (This comes with the caveats about how you don't want to be a slave to a single customer, etc. But we're having luck developing our core technology in the process for the big customer, and will re-use it for our product.)