Organizational agility developments
Many frameworks use the model ‘run the business’ (permanent teams doing the work) versus ‘change the business’ (work will be done by temporary groups of people). Projects and programs are managed via ‘change the business’. We see project and programme managers bringing people together for a definite period of time to make this happen. But in many cases we are confronted with budget overruns, delays and unhappy customers because what is delivered is not what they really need. As a reaction on these unsuccessful projects a group of people created the agile manifesto, and based on that several agile delivery frameworks were designed to help to deliver more successful projects.
This agile manifesto is embraced by many organizations and these organizations started to keep the people together who delivered e.g. a new system. These teams are able to deliver changes much faster by using Scrum, Kanban etc. These small focussed agile teams are self-organizing and are continuously learning to deliver more with the same people and within the same time-boxes. Collaboration is the norm. What these teams are delivering is managed by a product owner. It is the product owner who prioritizes the work to be done by maintaining a backlog with potential features and user stories. For these teams we don’t need a project manager to bring people together and structure and control the work. The members are already together, are self-organizing and can be seen as part of the ‘run the business’ side of the organization.
So for those changes that can be managed by a single agile team there is no need for a project manager. But in many occasions you need more than one agile team to implement the requested change. We need to scale up from a single agile team to many agile teams. The Scrum guide (developed by Ken Schwaber and Jeff Sutherland) gives some directions to use the scrum of scrums to align the teams and to make integration into one solution possible. In practice this works well for just a few teams, but what do you need if you have to make use of several component and feature teams to deliver one integrated solution? Scrum of scrums is not enough, you need a project manager to manage all stakeholders, all dependencies, the complexity and to deliver one integrated solution. Several organizations are on this level. They still run projects and project managers will make use of these permanent agile delivery teams.
Probable you could ask yourself, why do I want to make use of a temporary project manager? Is it not possible to have a permanent, maybe virtual, structure to take care of stakeholder management, dependency management and integration and have as a result a more or less continuous flow or at least short delivery cycles of changes bringing into production without ramping up and down project governance structures and teams.
You now see that several frameworks to support this level are being developed and used by many different organizations. To mention a few: Nexus developed by Ken Schwaber, Scrum at Scale developed by Jeff Sutherland, SAFe developed by Dean Leffingwell, the Spotify model copied by several organizations, Less developed by Craig Larman and Bas Vodde and there are many more. If you have implemented one of these frameworks the need for a project or programme manager will decline but on the other hand they can take new roles like roadmap manager, integration manager, release train engineer, value stream engineer.
Does this mean we don’t need any more project or programme managers? I think for the coming years we definitely need project and programme managers. In those cases, where we need more than the already existing permanent teams we have to build these non-existing teams. And these teams can of course make use of agile ways of working or just choose the for them most appropriate delivery method. If there is a need for a piece of specialist work we must select the right people and bring them together to deliver. This is typically a task for a project or programme manager. If you want to transform your organization, open new product/market combinations, integrate departments, or merge different organizations I expect that most of the organizations will not have permanent teams in place to handle this.
To support this way of working we see frameworks on project level (PRINCE2 Agile, DSDM AgilePM and PMI APM) and at programme level (MSP and DSDM AgilePgM). The competences of these project and programme managers have to change. where in the traditional way of working the focus was on project results using a formal mandate we now see a shift to business results realized by using influence without power. Stakeholder management becomes even more important with a focus on empathy, negotiations and persuasiveness. Servant leadership becomes key.
Here too, I see developments to reduce this type of project management work. Where we first saw integration of development and operations people into DevOps teams we now see the first BusDevOps multi-skilled teams where product management, marketing, development and ops people are in the same team and as a consequence again less projects and project managers.
Scrum or Kanban?
During the first part of a product life cycle the uncertainty is high and the focus is on goal driven iterations for the first product launch and market fit product. During this part of the life cycle Scrum is a great fit to cope with uncertainty and product iterations developed by the whole team. During the rest of the product life cycle the amount of uncertainty and change gradually declines. Here Kanban is a good fit. User Stories will be realized in a continuous flow by one or more of the individual team members.
When a major product upgrade has to be delivered by the whole team Scrum could be a better choice for that goal oriented iteration, otherwise Kanban stays a good fit.
To avoid the error prone handover and to shorten the time to market the Development and Operations teams can be integrated. Kanban is a good fit for the DevOps team. When to start with DevOps varies.
Agile portfolio management
Does this have consequences for portfolio management too? At this moment I have not seen any mature agile portfolio frameworks. The first framework that includes the portfolio level is the SAFe framework and DSDM included an agile portfolio management paragraph in their little book The Agile PMO.
In one of my previous posts I already proposed a change in the MoP framework to include the ‘run the business’ permanent agile teams in the portfolio view. If we want to reach more business agility, I strongly believe that we have to decentralize decision making. If we don’t and still want to make decisions at a higher more central level Standish ‘Cheetah’s Law’ is applicable and the speed of decision making could obstruct delivery success.
So for me the following aspects need to be taken into account to design an agile portfolio management framework:
Agile Portfolio Management Framework:
- Strategy assessment
- Internal and external environment assessment (SWAT)
- Portfolio management must facilitate sustainable business change (People, Planet, Prosperity, Processes, and Products)
- Strategic Objectives setting
- Develop Strategic themes
- Direction setting
- Portfolio vision, goals and objectives
- Portfolio management facilitates innovation as part of the roadmap
- Portfolio management must move away from the iron triangle and focus on delivering value, capacity and time-to-market
- Close cooperation between enterprise architecture and portfolio management (addressing enabler epics (NFRs, technology drivers, innovations) to be part of the roadmap) to invest in (digital) technology to win, serve and retain customers
- Portfolio management will have large impact on strategic decisions (achievability, technology trends)
- Funding at value streams or permanent agile teams level and not at project or programme level
- Funding must be aligned with the strategy or strategic themes. Enlarge or lower the number of agile teams must take place to align with the strategic themes
- A short simple business case justification must be used to put epics on a portfolio backlog
- The portfolio backlog epics must be prioritized based on attractiveness, risk or opportunity costs, time criticality and the duration. The weighted shortest jobs first (WSJF) from SAFe is a good example. Standish ‘Law of the eatable elephant’ is in line with this.
- Epics can be business related as well as non-functional
- Epics must be head and heart-driven, not just head-driven
- Keep epics as small as possible but it must contain more than one feature
- Number of epics in the roadmap must be WIP limited
- Portfolio plans will be replaced by a portfolio backlog with epics and a rolling-wave portfolio roadmap (Roadmaps include six key elements: time frame, prioritized and identifiable outcomes, strategic themes, context-specific content, dependencies, investment outlay)
- Starting point for a portfolio roadmap must be a portfolio vision
- Rolling-wave portfolio roadmap must be a living document. Only the first part must be committed to make sure changes can be embraced
- Portfolio roadmaps must have a cadence or heartbeat to increase throughput and integration moments/milestones to create learning loops
- Portfolio roadmaps must show retrospective events
- Portfolio roadmap achievability must be based on (group of) team(s) velocity and not on optimized resource utilization. 100% resource utilization will lead to a lot of busy persons but no delivery!
- Portfolio roadmap must be approved by senior management and communicated to the organization
- Must be a continuous integrated portfolio planning process with regular strategic reviews (included fact-based feedback loops) and pivot when needed
- Portfolio roadmap development includes strategic option analysis / scenario planning
- Portfolio dashboards must show the funding of value streams (and permanent agile teams) and the alignment with and budget allocation across the strategic themes
- Portfolio dashboards must show progress on epic level. Details of epic break downs in features and user stories are not for the portfolio level (respect the decentralized decision making)
- Focus must be on delivering value / benefits and not on OTOBOS (On Time, On Budget, On Scope)
- Possible portfolio dashboards Key performance indicators and metrics (not limitative): productivity (feature lead time), agility (predictability, number of releases), quality (satisfaction, #defects), metrics for self-improvement, time to market, NPS
- Use timely, accurate, and relevant information based on real time (automated) performance data, avoid manual aggregation
- Portfolio dashboards must show data-driven recommendations for decisions
- Portfolio dashboard reporting at anytime
- Dependency management on epic level (inter and intra dependencies)
- Doing the right things (metrics on effectiveness), Doing it right (metrics on process efficiency). Compare over more than one period
- Customer feedback to evaluate the effectiveness of the roadmap
- Portfolio dashboard reporting creates transparency and will motivate stakeholders
- Integrated tooling (EA and PPM) must give real time insights (rich information) about the health of initiatives, capacity and what-if scenario analysis corresponding with the requester’s role
- Senior management commitment (much more leadership, less management)
- Decentralized decision making
- End-to-end transparency
- Inspect regularly and adapt where needed
- Feedback is crucial
- Empowered employees
- Culture of collaboration (remove silo’s)
Looking forward to your comments and adjustments so we can co-create a new agile portfolio management framework.
- 30-09: added Scrum or KanBan paragraph
- 30-09: Agile Portfolio Management Framework additions
- 02-10 Picture Agile PfM Framework