Everyone wants their next big thing to be a hit. But entrepreneurs should know when to cut an idea loose, because it only takes one to go bankrupt. 

One good example is LEGO. In 1998, the company suffered its first loss in revenue. For a time, things were looking up: They'd just hired an idea man and were cranking out hits like the Star Wars and Harry Potter sets. But after launching a buildable action figure named Galidor, they moved away from the sets--and what customers had come to expect. Soon their sales tanked, expenses had soared, and by 2003, the company was on the verge of bankruptcy.

Michael Schrage, author and research fellow at MIT Sloan School's Center for Digital Business, writes in the Harvard Business Review about how to spot a bad idea before it gets too far. Here are two "pathologies" to watch out for: 

You're not gaining insight. 

Every innovation has a hypothesis. As you go along and test, your hypothesis should prove to be true. However, if your service or product isn't blowing you away with new findings or insights into value proposition or customer perception, bag it. "Each iteration and test--with and without customers--should be generating nuggets of data and information that facilitate a deeper understanding of what they're really doing instead of what they started out doing," Schrage writes. "Innovation journeys should, quite literally, change how they see their original ideas."

Schrage recalls sitting in a product meeting where a social media start-up was making a Facebook app. When the tests showed a lack of traction with customers, the product leader asked, "So what do we understand about our users now that we really didn’t before?" When no one responded, they killed the app. 

Your customers don't care.

Your product, service, or company should get people excited. And if you tweak your product and they still don't care, you're toast. "Ideally, they should come to care as much--or more--about your innovation as you do," Schrage says. "But if their feelings don't change as they engage more with your next iteration or simplified UX, you're in an emotional kill zone." On the flip side, if your customers dislike a new prototype that could mean they liked what you were doing before. But if they disliked the beta version, chances are they'll dislike the official one.