The expensive ideas are not the bad ones. Bad ones die on their own. The expensive ones are good enough to start and never explicitly decided on.
You had the idea in the truck, or in the shower, or at 11pm. It was good. You told someone about it and they were interested, which felt like confirmation. You started sketching it out. Then you started building it.
Three weeks later you are deep enough that stopping feels like waste, and you still have not answered the question you never asked: is this worth doing, compared to the other things you could be doing with three weeks?
Nobody ever said yes to this project. Nobody said no either. It just accumulated momentum, and momentum is not a decision.
Three acts that get collapsed into one
The reason this happens is that thinking about an idea and committing to an idea feel like the same activity while you are doing them. They are not.
Capture
Get it out of your head, fast and messy, without judging it. Judging at capture time kills good ideas and, worse, makes you stop capturing.
Shape
Turn the thought into something examinable. What would have to be true for this to work? What does it cost? Who is it for? What are you assuming without evidence? This is where most ideas quietly reveal themselves.
Authorize
A separate, explicit yes or no, made by a person on a specific day. Not implied by having started. Not implied by enthusiasm.
The assumptions step is the valuable one
Most ideas contain one load-bearing assumption that has never been said out loud. Customers will pay more for this. The extra work fits in existing hours. The people asking about it would actually buy it.
Written down as a question, that assumption is usually cheap to check, and checking it takes an afternoon instead of three weeks. The workflow does not answer it for you. It makes sure it gets asked before you commit.
The part that is worth saying plainly
This will kill some ideas you liked. That is the point, and it is the return. A no on day one costs an hour. A no on week three costs three weeks, and it usually does not arrive as a no at all. It arrives as the thing quietly getting abandoned, which teaches you nothing.
What it does not do
It does not evaluate your idea. No workflow and no model knows your market, your capacity, or your appetite for risk. It structures the thinking and forces the decision to be made by a person, on purpose, at a moment you can point to later.