Did you fail fast?
I came across this Ted talk for Google X - arguably one of the most innovative companies in our current time that puts most crazy ideas in realisation and gives employees bonu$ to fail.
Simply put, X first runs the most difficult bits of an idea through the following process: trial & error --> data --> analyse --> fail/proceed. Same process will repeat for next element of the idea if final decision is a go. Each iteration is short and has a complete feedback loop to assess the alignment and opportunities, e.g., the change of perspective on self-driving car example.
Much like the agile development methodology, if the idea is architected in a way that is loosely coupled and less dependent, several process iterations may even run in parallel, which ultimately reduces the time-to-market or time-to-fail.
In all, the culture of fail fast is certainly the cornerstone of breeding an innovative culture. To be sustainable and not go bankrupt on the first moonshot idea, fail fast should then be elaborated to controlled failure and fail small. That is, setting the right context and slim-fit boundaries together with a firm management commitment (e.g., bonus on employees who fail a project early).
Often the most challenging aspect is the management commitment. As a manager, you would do the risk assessment and then gauge whether promoting "fail fast" will lead to an innovative culture and if yes, how long or how many failures will it take? - it sounds like a good first candidate for "fail fast" trial run.