You have to put yourself into a position where the team will know if it's working or not working and the problem is that most teams have a plan which is basically ship it and see what happens. And the problem with ship it and see what happens is that you can feel like you are very agile but you are guranteed to succeed at seeing what happens and so you therefore always will feel like it is worth doing and you are feel like you are on the right track no matter what.

The only way to get yourself into a position where you have to pivot is to make concrete specific predictions ahead of time, that if they turn out to be wrong will actually call your theory into doubt.

And the issue is that we all know that most projections for new products are complete BS. You have to tell the CFO or whoever that you have a zillion tillion customers in year five like otherwise you won't get the money to do your project but we know that we just made those projections up so when they don't prove to be correct we are like "oh that doesn't prove that our vision is wrong, it just took longer than we expected".

Eric Ries in "The Lean Startup" | Talks at Google