I have worked with a lot of people, teams and organizations over the last 8 years helping them to adopt Scrum and I have seen some interesting patterns about where Scrum works well and where it doesn’t work so well. I wanted to share my observations to see if they correlate with what other people are experiencing.
So first off, I want to describe what I mean by Scrum working well:
- Teams using Scrum are obviously high-performance teams whose business results are at least 4x that of normal teams.
- The organization in which Scrum is being used experiences a change in culture to become more team oriented, more value oriented, and more customer oriented.
So now I can describe where I have observed Scrum to work really well:
- When an organization (or team) is in deep trouble and willing to admit it adopting Scrum seems to be a catalyst for creating a new culture, process and team environment where getting out of trouble is possible. This is Scrum for Crisis. The “willing to admit it” part is extremely important as I have worked with two organizations where the “deep trouble” part was obvious to me as an external person, but in both cases management and staff did not seem willing to admit the depth of their crisis and in both cases Scrum failed to act as a catalyst to get them out of trouble. In this use of Scrum, sometimes resolving the crisis then leads back to complacency and Scrum fades away.
- Small growing organizations that have no existing formal processes for development can use Scrum as an effective way to maintain their high-performance without getting burdened in bureaucracy. In this case, it is important to note that they are already in a high performance state and their struggle is to maintain that while at the same time growing. I’ve worked with quite a number of small organizations where all they need is the CSM (plus maybe one or two days of coaching) to adopt and maintain Scrum. I have also worked with small organizations that were not already high performance and Scrum has not typically worked to bump them up to a high-performance state.
- Pure new product development where a single strong Product Owner can be identified who has the authority to make product decisions independently of anyone else (including product budget decisions). By “pure new product development” I mean that neither the individual team members nor the team as a whole have any responsibilities outside of the product work – there is no “fractional allocation” or “resource levelling” across projects or products. The strong Product Owner is critical to success with Scrum and must understand the principles of Scrum as well as the mechanics of being a Product Owner.
I have also seen Scrum be inappropriate and not lead to the results I mentioned above:
- Management teams. It seems like Scrum could or should work for management teams, but it appears that managers have too much of the following problems to be able to use Scrum:
– operational responsibilities (non-creative, non-problem-solving work)
– urgent, legitimate interruptions (e.g. an escalated customer issue)
– real commitments to events or projects that are calendar based (e.g. a management off-site)
– ego: they don’t want to follow an apparently rigid process or they are always happy to make exceptions for themselves
Again, one might imagine that Scrum should work to help resolve these issues, but unfortunately I have never seen it able to do so in this context.
- Small teams/projects. Scrum is too heavy for teams less than 5 people or for projects shorter than 2 months in total duration. Those numbers aren’t meant to be hard and fast, but when I’ve seen small teams/projects attempt to do Scrum they always end up breaking lots of the rules partly because they can and partly because they must. That said, some folks have created “Personal Scrum” and other variants. I’m not sure if we as the Scrum Alliance officially recognize/endorse those variants.
- Purely operational work. There just isn’t enough creativity/problem-solving to make the Sprint an appropriate process element, nor the Product Backlog an appropriate organizing mechanism. I have seen some operational environments get some benefit from doing regular retrospectives, but just doing retrospectives is not Scrum in my book. My experience with Kanban is still a little limited, but it seems to be an appropriate approach for these environments.
- Organizations where there is very little need to change. I’ve spent some time working with big profitable banks to adopt agile and without exception, they just can’t wrap their minds around the need for Scrum… because they are already so successful as a business. The general attitude is that Scrum is popular therefore we will call what we are doing “Scrum”, but it really isn’t. It’s Scrummerfall and Scrum-Butt wrapped up in the terminology of Scrum. They will adopt some Agile practices and get very modest benefits. I have seen minor improvements in team morale and minor improvements in quality and productivity, but certainly not anything near to what is possible for improvements. When we do assessments in this type of environment, we see Value Stream maps with waste at the 80-90% level so there is huge room for improvement… but it just doesn’t happen.
Scrum can definitely transform the world of product development. It can definitely act as a catalyst to get teams and organizations out of crisis. But that isn’t the whole world of work. I’m also concerned about the idea of using Scrum for general project management. There might be some good practices that are part of Scrum that would also be valuable in general project management (e.g. regular retrospectives, daily team meetings) but that doesn’t make Scrum a general project management framework.
I don’t claim that any of the above observations are “correct”. That’s partly why I am sharing – I would love to have a good discussion here about this because I think it is critical for us as Agilists to be able to answer this question well when we are asked: “what is Scrum good for?” – particularly since Scrum is by far the most popular Agile method.
I would love to hear other people’s observations about where Scrum works well (as I have defined “well” above) vs. where it either is only a modest improvement to existing approaches or where it might even not work at all.