Going too far is the only way to know you've gone as far as you could
It used to be that banks rewarded loan managers according to the number of customers who made good on their loan repayments: the fewer customers defaulted on payments, the higher the end-of-year bonus. So very quickly loan officers figured out that only approving zero-risk customers would yield the biggest bonus. And it worked: all the loans were repaid and the loan managers got their big bonuses.
But the banks weren't happy: they make money on interest fees and by restricting the loans to a small number of customers they were missing out on revenue from all the other customers who might pay back their loans - but weren't given the chance. By eliminating all risk, the loan managers were hampering profit.
So the bonus system was changed and loan managers would get the maximum bonus even if 5% of customers defaulted on their loan. By including a small loss in the measure of success, the banks ensured that all the potential customers had been included. It would mean failing a little, but that would just be proof they were lending to all the customers they wanted to target.
If you never fail you're playing it too safe and you'll never know how much further you could go. By playing it safe you might be missing out on so much more: ideas, processes, people ... always ask yourself: Why stop there? What happens if I carry on? Where does it break?
To find the limit you have to go a little too far and then bring it back in: controlled failure is the only way to know you've gone as far as you could.