What's not importantMany organizations start by defining their future structure and ensuring that all important people have gotten an "agile" role. Let me be blunt: in many cases, the transformation has already failed at this point, because it will be the same people doing the same thing in the same way as before, just with a new label.
So then, what is important?
#1 Relentless improvementMany organizations tailor the Emperor's new Clothes. They set up systems where transparent, honest criticism isn't welcome - only good news and affirmation of status quo are acceptable. Even when they do indeed set up ceremonial Inspect+Adapt events, these events will never result in fundamental change and are usually just affirmation sessions leading to cosmetic changes when invasive surgery would be required.
To become agile, the organization must be able to accurately identify where things are going wrong, then precisely locate what aspect of the system is the root cause, then mercilessly make effective changes. And then, they must acutely observe the outcome of that change. If this isn't happening on a frequent basis, tranformation just establishes a new status quo.
An organization without a culture of Relentless Improvement will be cargoistic - they will go through all the motions in a ritualistic way and never understand why they're not seeing the desired benefits!
A) Systemic Inspect and AdaptTeam Retrospectives are just a small part of the picture. Everyone should be encouraged to identify and highlight systemic dysfunctions. The organization must have a transparent process for addressing, analyzing and resolving systemic dysfunctions. This takes courage on everyone's side - and openness from leadership.
Agile coaches, respectively Scrum Masters, first and foremost, need to work on creating both this kind of culture and awareness.
When effective systemic Inspect and Adapt is in place, most of the other things will eventually fall into place.
B) Closed feedback cyclesFeedback occurs everywhere, anytime: Among individuals, during the work, in processes ... If feedback is accepted and acted upon, further feedback is encouraged. When feedback is routinely rejected, the cycle breaks. And when no mechanism for providing feedback exists, the door to improvement is closed.
Agile coaches need to keenly observe how feedback works and actively shape closed feedback cycles.
C) Radical CandorWhen the Emperor has no clothes, people must be able to say so. A dysfunctional process must be fixed, a worthless requirment trashed and an undesirable feature axed. As long as there's a mindset that decision makers know why they want things and the system doesn't incentivize stopping the wrong things, the most significant improvements to performance and outcomes just won't happen.
Agile coaches need to lead by example in being candid: Questioning worthless work and dysfunctional structures are bare minimum requirements to effective change.
#2 - A strong product organizationLet me be clear, I don't mean that a lot of people should be involved in "Product Owner/Manager" roles. A product organization is strong when it has effective mechanisms of addressing the right things and setting developers to do the right thing. The product organization isn't as much about structure as it is about turning opportunities into value in a very short time. It's about setting up the mechanisms and instilling the knowledge for maximizing value.
A) Focus on ValueProduct people need an appropriate, effective way of discovering where value lies, identifying which backlog items have how much value and extracting the primary value from larger topics. When product people aren't keen on maximizing the value delivered, all is lost. Any mechanisms or incentives within the organization which tint, bias or distort the understanding of value will multiply a hundredfold into ineffective outcomes, so they need to be addressed and removed.
Agile coaches need to establish a focus on value by helping product people and managers see what is value and what isn't.
B) Deferred Decision MakingUntil it's clear whether a value hypothesis is even valid, product people should reserve the right to change direction. When items are scoped for implementation before it's clear that their delivery will generate any value, there's a lot of money at stake. Should exploratory work open up alternative routes, the product organization must be able to inspect and adapt to this!
#3 - Technical ExcellenceDoing the right thing in the wrong way may make the outcome appear like it was the wrong thing to begin with. Shoddy workmanship costs money, wastes opportunities and causes further harm to outcomes in the future. Developers need to be able to take pride in a job well done and have the confidence that their results are of high quality.
A) Genuine Shift-LeftDevelopers can learn how to better build products which do what they're supposed to in a way that customers like. Having a separate Testing or Operations unit is counter to this learning taking place. Terminology like "DevOps Team" masks dysfunctions and fails to address the root cause.
Agile coaches need to spot where functional separation breaks feedback cycles and bring the right people together.