We know that some innovations work and some don’t. We are often told to admit it, learn from it, and move on. But do we always do this? And can we?

Too often when we think of innovation we think of new ideas, and far too often there is an assumption that new means success. Yet this is not true, experimentation brings risks, and some things will inevitably not go to plan. There needs to be experimentation and data to be acted on. There also needs to be consideration about how much failure is a failure in itself.

The recent book, Blunders of Government, outlined a number of examples to illustrate that good intentions don’t always lead to good outcomes. Often failure can be costly financially, socially, economically, and to individual reputations.

There are also frequent accounts in the media of unacceptable failures to services, from issues in hospitals, to children let down by social services. What then is an acceptable level of risk and failure?

Our recent i-teams research analysed 20 innovation teams, funds and units across six continents and revealed that attitudes to failure vary greatly across the globe. It seems that USA i-teams are most at ease in recognising failure, actively discussing how much is acceptable and how they handle risks.

In Europe, some innovation teams and units discuss the difficulties of managing risk, but many don’t explicitly discuss the potential of failure. Whilst in Asia and Latin America, risk is often said to be completely unacceptable.

The boldest acceptance of failure we found in our research was in New York City. When Mayor Bloomberg launched the NYC Center for Economic Opportunity, he acknowledged that a number of projects would fail, providing explicit political cover for the team to take risks. One NYC CEO staffer noted, “if everything you do succeeds, it probably wasn’t risky enough”.

It is often argued that a team or organisation's relationship with, or proximity to, government can impact on how radical it can be, with those closer to politics being more constrained. Yet paradoxically some of the most radical innovation teams we came across – like NYC CEO - are based in the mayor’s or president’s office. Rather than hindering experimentation, this closeness to political influence actually enables risk taking to happen.

What is an acceptable level of risk is not just defined by government leaders. The public also have a say. When in Seoul last year I spoke to a government official who said that city hall is there to deliver services for citizens, not to run services which fail.

Now there may be exceptions to this rule, we know that failure and success are not binary, and this analysis is admittedly based on a small sample, but understanding a country’s attitude to failure does throw up interesting questions about how much risk government should and could take.

Innovation teams and their wider government need to be frank about how much failure is acceptable, both across the board and in specific policy areas. This could shape whether their innovation agenda is one of more incremental improvements to existing services, or radical overhaul.

Neither is necessarily better than the other, both can lead to marked improvements in outcomes, but the choice will influence the methods, approaches, skills, and resources required.

There is also a wider point here about impact measurement. NYC CEO describe their impact data as their “first line of defence” when experimenting with new services.

Yet this is rare, too often data is lacking, which means programmes and policies are probably failing all the time, we just don’t have the ability to recognise it.