Scrum@Scale
Scrum@Scale is Jeff Sutherland's scaling framework, built on two linked cycles — one for what gets built, one for how — that repeat fractally at every level. It is modular rather than configured: you adopt the components your bottleneck requires and leave the rest alone. This page lists every component of both cycles, the two points where they touch, and the roles and events each one adds.
- Author
- Jeff Sutherland
- Scaling unit
- Scrum of Scrums
- Group size
- 4–5 teams
Two cycles, repeated fractally
Scrum@Scale splits the work of running an organisation into two separate cycles. The Scrum Master cycle owns how work gets done — impediments, improvement, coordination between teams, getting things into production. The Product Owner cycle owns what gets done — vision, backlog order, decomposition, release decisions.
Keeping them separate is the design point, not an accident of the diagram. In most organisations the "how" side quietly takes over the "what" side: delivery managers end up deciding priorities because they control capacity. Two cycles with their own hubs, meeting only at defined points, is the structural answer to that.
The second idea is that the framework is scale-free. Four to five teams form a Scrum of Scrums, which behaves exactly like a Scrum team: it has its own daily, its own backlog of impediments and its own goal. Four to five of those form a Scrum of Scrum of Scrums. Nothing new is invented at each level — the same pattern repeats, which is why an organisation does not have to learn a different set of practices as it grows.
The five-team grouping is not arbitrary — it is the same span-of-communication constraint that caps a Scrum team at around nine people. Above that, the daily coordination meeting stops being a conversation and becomes a broadcast, and the people who no longer speak start solving their coordination problems in private channels instead.
The Scrum Master cycle: the "how"
Three components sit on this cycle. Each is a capability the organisation either has or does not — naming them separately is what lets you say "our problem is the third one" instead of "we should do more agile".
Continuous improvement
Identifying impediments and actually removing them — including the ones a team cannot remove itself. The measurable test is whether the same impediment appears in two consecutive retrospectives.
Cross-team coordination
Aligning the work of teams that share a goal, mostly by removing dependencies rather than by managing them. This is what the Scaled Daily Scrum exists for.
Delivery
Getting an integrated, releasable product into users' hands — the pipeline, the environments, the release mechanics. Earlier editions of the guide call this component Deployment.
The cycle's hub is the Executive Action Team, or EAT. It is a real Scrum team made up of leaders with the authority to change the organisation, and its backlog is the organisation's impediments. That authority requirement is the whole point: an impediment like "procurement takes six weeks to approve a test environment" cannot be removed by a Scrum Master, and if nobody with budget authority owns it, it is raised every sprint forever.
The role on this cycle is the Scrum of Scrums Master, or SoSM. They are accountable for the release of the joint work of their group of teams, the same way a Scrum Master is accountable for their team's process. The scaled events are the Scaled Daily Scrum — a short daily where representatives surface cross-team blockers, held after the team dailies — and a scaled retrospective where the group inspects itself as a unit.
The Product Owner cycle: the "what"
Four components sit on this cycle. Together they answer the question of how one direction survives contact with twenty teams without a project management office deciding it for them.
| Component | What it answers |
|---|---|
| Strategic vision | Why this product exists and where it is going, stated compactly enough that a team can use it to make a decision without asking. |
| Backlog prioritisation | What gets built first when every stakeholder's request is urgent. One ordered list, decided by people with the authority to say no. |
| Backlog decomposition and refinement | How a large piece of value becomes items small enough for a team to finish inside one sprint, with dependencies between teams reduced. |
| Release planning | What is expected to be released and roughly when — a forecast built from actual delivery data, not a commitment made before the work is understood. |
The cycle's hub is the Executive MetaScrum, or EMS — the forum where every competing stakeholder priority is reconciled into one ordered backlog by people with the authority to decide. It meets on a regular cadence, usually at least once per sprint, and it is run as an event called the MetaScrum. The role that leads this cycle is the Chief Product Owner, who owns the ordering of the whole product backlog the way a Product Owner owns one team's.
Where the two cycles touch
The cycles share exactly two components, and everything else runs independently. Those two shared points are where the "what" and the "how" have to agree, and limiting them to two is what stops one side absorbing the other.
Team-level process
- Where a prioritised backlog meets a team that can build it
- Sprint Planning, the sprint itself, the review
- The PO decides what; the team decides how much and how
- Unchanged from ordinary Scrum, deliberately
Product release and feedback
- Where built work reaches users and returns as evidence
- The delivery capability meets the release decision
- Feedback re-enters the PO cycle as backlog order
- Without it the two cycles spin without learning
It is worth being blunt about the second one. If releases are quarterly and feedback arrives as a report three weeks later, the Product Owner cycle is prioritising on opinion. Every other component can be in place and the organisation will still be guessing — which is why the delivery component on the Scrum Master side is usually the first thing to fix, even though it looks like a purely technical concern.
Scrum@Scale vs. SAFe
Scrum@Scale
- Modular: adopt components against a diagnosed bottleneck
- Scale-free — the same pattern at every level
- Keeps the sprint as the planning horizon
- Light on new roles; reuses Scrum accountabilities
SAFe
- Configured: pick a named configuration and adopt it
- Distinct practices at team, program and portfolio levels
- Adds a Program Increment of 8–12 weeks
- Introduces several roles above the team
The practical trade-off is where the judgement has to come from. SAFe carries the judgement in the framework — the configuration tells you what to stand up. Scrum@Scale expects the organisation to supply it, which is cheaper and lighter when the diagnosis is right and produces a half-built set of meetings when it is wrong.
How this shows up in real delivery
Scrum@Scale's modularity is its strength and its risk. Adopted well, an organisation diagnoses that its bottleneck is prioritisation, stands up an Executive MetaScrum and fixes it without touching anything else. Adopted poorly, "modular" becomes permission to take the meetings and skip the accountability — a Scrum of Scrums that reports status upward and removes no impediments.
Where it degrades
- A Scrum of Scrums used as a status meeting rather than as a team with its own impediment backlog.
- No Executive Action Team, so organisation-level impediments are raised forever and never removed.
- Skipping the Executive MetaScrum, which leaves prioritisation to escalation.
- Groupings far larger than five teams, where the coordination meeting stops being a conversation.
- An Executive Action Team made of delegates without budget authority, which can name impediments but not remove them.
- The two cycles merged into one management forum, after which the "how" side starts deciding the "what".
When to use it
Use it when
- Teams already run Scrum well and you want to scale the same pattern rather than learn a new one.
- You have diagnosed a specific bottleneck and want to address only that.
- Prioritisation across stakeholders is the real problem — the Executive MetaScrum targets it directly.
- You want to keep the sprint as the planning horizon at every level.
Avoid it when
- Teams are not yet running Scrum competently — scaling a broken pattern scales the breakage.
- The organisation wants a prescribed end state rather than a diagnostic approach.
- Leadership will not stand up an Executive Action Team with real authority.
- Only integration is broken and there are fewer than nine teams — Nexus is a smaller fix.
Found this useful?
Share it with someone who is working on the same problem.