Personally, I am not prepared to believe that Scrum's failure rate is because people don't implement Scrum properly. That's the whole Scrum-but argument. Scrum is such a light framework, if you give it a serious go, it's pretty hard to really do it wrong. That said, I do believe that many organizations aren't prepared to fix the organizational disfunction that Scrum is going to show them. Scrum's primary benefit is that it shows you the stuff you need to fix. If you aren't willing to fix the problems, you aren't going to get the benefit.
How many companies using Scrum are really trying to transform the entire enterprise? I'd wager that most Scrum teams live within a larger, more traditional enterprise. In these organizations, structure and culture are going to trump anything that a single Scrum team does with people, process, or tools. You could have the best team members, run a perfect Scrum, have big visible charts everywhere, and I bet you still stand a chance of falling into Schwaber's 75% statistic.
For everything we do to build a high-performing agile team, the organization is going to work to pull that team a part. The pull might be intentional, led by people that are threatened by any change to the status-qou. More than likely, the organization just doesn't get it. The value delivered by the Scrum team just get's lost within the larger underperforming organization. Even though the team was effective, it just never really translated to any kind of measurable value for the larger organization.
So therein lies the problem. What does it mean to have a successful Scrum team in an organization that isn't performing very well? What do you do when Scrum helps you identify an impediment that can't be solved by the team, but no one outside the team seems to care? Scrum ends up being a local optimization within the larger enterprise value stream. Unfortunately, it doesn't matter how hyper-productive your team becomes... getting higher team velocity isn't your biggest problem.
The trick to adopting agile is to form a team around a business problem the organization actually cares about solving... one that will increase bottom line performance. And this is where the conversation get's interesting. Many of us aren't responsible for improving the whole system. We are only empowered to do our part. We can only operate within our circle of concern. Our challenge is to figure out a way to make our local improvements, but in a way that is respectful of the whole.
There are a few things I want to suggest that will help you understand how your company looks at value delivery. For the rest of this post, I want to talk about aligning your agile pilot team to something your organization actually cares about. The important thing to realize is that not every company has the same kind of goals. If we are going hook ourselves up to big business problems, we need to understand how our companies look at value... how they look at the unit of production. For most, a unit of production is not a single user story delivered by a single team.