Recently I was at a Scrum discussion group hosted by Tobias Mayer. Noel Warnell posed the question, “Why scale?”. That was about as clickbaity as the title to this article and certainly attracted attention; and the conversation led me to the following realisation.
When we talk about scaling, we fall into one of two camps: either we are looking to grow our delivery capability and realise we will have too many people for a single team, or we already have lots of teams and are looking to make their delivery better. These two situations are enormously different and SAFe is suitable for one, but not the other.
When there is only a single team, likely collocated and cross-functional, then things are probably working well. The relationships within the team will be strong enough as to not require any compensating processes and you’ll find the Scrum guidelines will be a fairly comfortable fit. With this situation, if you need to grow, then your focus will be to not break things. You want to increase delivery rate BUT maintain all the great stuff you have with that engaged team.
In contrast, consider a situation with lots of teams all working together, in a borderline anarchic system, chances are things have grown organically or possibly as a result of an acquisition. I would expect multiple locally optimised groups, possibly clustered around technical components that are drowning in dependencies and technical debt. Having seen this many times and find it often stems from a misalignment of objectives and a “too busy to improve” culture. Commonly there is failure to appreciate it is attitude and relationships, not individual talent, that make great delivery. In this situation the objective is to fix the system rather than grow it.
When we talk about scaling we need to be clear whether we mean to GROW something, or FIX something, and the solutions to that challenge are likely to be very different. This brings me to SAFe.
Returning to that single team, all keen and dynamic and you put the SAFe picture on the wall and proclaim proudly that one day we’ll be a huge machine all using this approach, chances are you’ll get some strong feedback. There are very clear and understandable reasons why not a single signatory to the Agile Manifesto has endorsed SAFe; it isn’t really very Agile in the original context (and for more context and a few rants, Google is your friend). What you’d probably be looking for is something more like a hive mind, a network of highly engaged teams each swarming over their own challenges, connecting over common values and simple clean components. Pretty sure that isn’t SAFe.
However in the other situation, if you have an anarchic system with a lot of effort but little value, high utilisation but little alignment, then the introduction of something with significant structure and assurance for those that are trying to make sense of the madness is not the worst thing to do.
We need to consider how we naturally would look to improve a complex system. I suggest we’d take the following steps:
- Impose order to understand the current state
- Identify opportunities to divide the system into simpler parts
- Ensure a simple interface between potential parts to ensure they can separate without immediately rejoining or breaking
- Reorganise to break the system into smaller parts
- Improve each part in a custom fashion for what you need it to be
If you consider that Scrum is a delivery framework for solving complex problems, then this approach isn’t too far from that. Apply order (sprint), break work down with clear interfaces, deliver…
Realistically if you discussed the situation with senior management, the chances are they would like to reach stage five for their teams but are very nervous as to how to achieve it. The challenge is always to be able to improve whilst maintaining delivery expectations; change the plane’s engine while keeping it in the air. Very crudely, the LeSS framework looks to get to stage 5, but rather optimistically looks to leap to step 4 and hope that through strength of character things will pan out well. SAFe, on the other hand, is a lot less disruptive and so compliant to the status-quo governance and culture, that it really doesn’t have ambitions beyond the step one above.
If you are on a building site and it is on fire, it doesn’t matter how spectacular the architect’s plans are, what you need right now is a fire extinguisher. That is how I see SAFe: it is a coping mechanism. Once the fire is out, then you can start building. Awareness is a critical prerequisite to change, and SAFe brings that clarity, however it won’t really fix the situation. SAFe also comes at significant cost to surface and manage all those dependencies, and doesn’t really suggest how to remove them, but once you have visibility of what is happening, what value each role adds, or could add, then you can appreciate how to improve further. The real benefits come from moving beyond SAFe, not moving onto it.
So SAFe is safe as long as what you have is already broken, but only remains safe if you don’t stay SAFe forever.
Comments are welcomed – even criticism. It is only through feedback that we learn.
Follow me on twitter: @philagiledesign