
For example, take a transformation that renamed 12 project teams as squads. Each squad got a backlog and a sprint, the weekly status meeting became the stand-up, and the sprint reviews went on one shared calendar. The work was still funded from 3 department budgets, each set once a year, the same way the projects had been. The teams got new names; who funds the work and who decides what it's for stayed exactly where they were.
Nothing in that story is a failure of the events. The stand-ups ran, the retrospectives ran, and everyone used the new words. That is what makes shallow agile scaling hard to see: from the conference room where the rollout is presented, a scaled framework and a scaled organization can look identical.
Mechanics and structure are two different lists
I separate what changed in a scaled unit into two lists. Mechanics are the parts you can install: team names, event calendars, roles, boards, and the tool everyone logs into. Teams need those things, and none of this argues they don't matter. Structure is who funds the work, who owns an outcome end to end, and who can decide without asking another silo.
I've seen everything on the first list change inside a quarter, and some of it you can buy off the shelf. The second list is harder, because changing it moves power. Someone gives up a budget line, someone loses a sign-off, and in a traditional corporate hierarchy that is a long conversation. So a transformation can change every item on the mechanics list and leave the structure list untouched, and a rollout report can still say it scaled.
It's the same failure we described in code a few weeks ago, where copying code instead of restructuring it left a team with debt it never booked. Copy the surface, skip the hard restructuring, and you end up with debt. At the level of the organization, the debt is a silo with new vocabulary on top of it.
What Digital.ai's 2025 survey measured
The numbers this week come from Digital.ai's 2025 State of Agile survey, its 18th edition. It is vendor-published: Digital.ai sells agile planning software. 349 Agile practitioners answered it in July and August 2025, each describing their own organization, and every figure below is a share of those respondents. In the same survey, 32% of respondents worked in organizations of more than 20,000 employees, and 56% were in North America.
Start with how deep Agile runs. In the same survey, 13% said Agile is deeply embedded across business, technology and supporting functions, with work aligned to strategic outcomes. That is a high bar, three conditions in one answer, and I think part of why the share is low is the wording. The other answers spread out: 25% said Agile is practiced across some departments but inconsistently, 22% scaled across IT with limited business involvement, and 20% mostly limited to team-level execution.
The same survey asked about investment. In Digital.ai's 2025 State of Agile survey (18th edition, vendor-published, 349 respondents), 41% said their organization's investment in scaling Agile increased over the past 12 to 24 months; 24% said it decreased and 27% said it stayed about the same. So more respondents reported an increase than a decrease, 41% against 24%, and nearly a quarter went the other way.
Those are two separate questions. The report publishes no cross-tabulation of them, so it can't show whether the respondents who reported more investment are the ones who said Agile is deeply embedded, and the 13% is never a share of the 41%. Side by side, though, they frame the question this guide is built around: when a unit scales, did its structure change, or only its mechanics?
Digital.ai, 18th State of Agile Report, 2025 (vendor-published): the report as Digital.ai publishes it
Where the investment goes is the part the survey didn't ask
The survey doesn't ask what the investment bought. The nearest it comes is an answer label: respondents who said their investment increased somewhat chose the option that reads "We've expanded Agile practices or tools." A label is not a measure of what the investment bought, so read it as a hint and no more.
The survey stops there. What I've seen scaling investment buy, more often than not, is mechanics: new team names, a shared event calendar, a new tool and training on how to use it, rolled out evenly across silos that stay where they were. That standardizes the vocabulary across the silos. It doesn't give them a common budget or a shared outcome, and I've come to think that buying the mechanics without the structure just adds overhead.
That's why I've learned to start with funding and decisions from the beginning, and with how teams are grouped around the stakeholders they serve. The mechanics will be there for everyone to see either way. Skip the structure at the start and the cost shows up much later, as a team that can't stop building the wrong thing.
It isn't about which framework you run
You might be thinking your organization doesn't run a named framework anymore, so none of this applies. The survey shows plenty of variety. Asked which frameworks and delivery models their organization uses, 48% of respondents chose a blended or hybrid model, 44% SAFe and 26% a homegrown framework or operating model; respondents could choose more than one, so those numbers don't add up. We've cited the hybrid and homegrown figures before, and they make the same point here.
This isn't a ranking, and the framework argument isn't one I'm interested in having again. A hybrid model can standardize the mechanics as well as a named framework can, and a homegrown playbook can rename the teams too. The question isn't which model you picked; it's whether the model touched funding, ownership and decision rights, or only the vocabulary. If your problem is the opposite one, a framework that has piled on process until it became the drag, that's a different diagnosis, and we covered the signs a scaling framework has become the bottleneck.
Where the silo still lives
If the structure didn't move, you should be able to find where it is still holding on. The survey gets close in one question, which asked what role business and executive leaders play in guiding Agile practices. In the same survey, 33% of respondents said Agile is treated as a delivery function, not a strategic lever, and 15% said business and executive leaders are actively involved in shaping and sustaining Agile practices; we've cited that 15% before. Another 24% said Agile is driven from the bottom up with minimal executive involvement.
Remember who answered: Agile practitioners describing their leaders, not leaders describing themselves. And notice what the 24% describes: minimal executive involvement in the practice, not leaders directing it. As I read it, staying out of the practice isn't staying out of the money. Leaders can leave the events to the teams and keep the budget and the big decisions for themselves, and that is the pattern I'd look for.
Go back to the 12 squads, funded from 3 department budgets, and say one of them owns the checkout experience. Its backlog says so and its sprint reviews show checkout work. But the money for checkout features sits in the payments department's annual budget, and that department's head decides whether the checkout plan changes. On paper the squad owns checkout; in practice a silo still owns the budget, and the squad depends on it.
We made the same point about AI pilots: ownership and decision rights are both on the list a pilot has to clear before it graduates to production.
The Silo Test
You can't fix a structure you haven't separated from its mechanics, so the Silo Test starts by separating them. Run it on one unit: a team, a group of teams, or a whole product grouping. It takes one sitting and one honest list.
List everything that changed when the unit was scaled, sort each change into two columns, Mechanics and Structure, and then ask three questions about the unit, whatever landed in the Structure column. Each question gets one of two answers: changed, or nothing changed. Read the columns side by side at the end. If every change sits under Mechanics, the unit was renamed, not restructured, and that is a starting point, not a verdict on anyone.
THE SILO TEST
One unit: a team, a group of teams, or a whole
product grouping. One sitting.
UNIT: _______________________________________
1. LIST EVERYTHING THAT CHANGED WHEN IT SCALED
Names, events, roles, tools, budgets,
approvals. All of it.
____________________________________________
2. SORT EACH CHANGE INTO TWO COLUMNS
MECHANICS: the parts you can install.
____________________________________________
STRUCTURE: who funds the work, who owns the
outcome, who decides.
____________________________________________
3. ASK THREE QUESTIONS ABOUT THE UNIT
Answer each: CHANGED or NOTHING CHANGED.
FUNDING: has the way the unit's work is
funded changed since it scaled?
Answer: _____________________
OWNERSHIP: has the team come to own an
outcome end to end, or does the work still
pass between departments?
Answer: _____________________
DECISION RIGHTS: has the team gained the
right to change what it builds without a
sign-off from outside?
Answer: _____________________
4. READ THE TWO COLUMNS SIDE BY SIDE
Every change under MECHANICS means the unit
was renamed, not restructured. A starting
point, not a verdict on anyone.
Then pick one structural answer to change
first, and name who would have to agree.
Change first: ______________________________
Who must agree: ____________________________
DONE WHEN: every change sits in one column, all
three questions have an answer, and one
structural change has the names of the people
who would have to agree to it.The worked example below runs the test on the checkout squad, so you can see each step filled in before you try it on your own unit.
FOR EXAMPLE.
The checkout squad, one of 12 squads funded
from 3 department budgets.
MECHANICS
Squad name, backlog, sprints, stand-up,
sprint reviews on the shared cadence.
STRUCTURE
Nothing. The column is empty.
FUNDING NOTHING CHANGED
Checkout money sits in the payments
department's annual budget.
OWNERSHIP NOTHING CHANGED
The squad owns checkout on its backlog;
the plan belongs to payments.
DECISION RIGHTS NOTHING CHANGED
Changing the checkout plan takes the
payments department head's sign-off.
All three answers point at payments. That
is where the conversation starts.
CHANGE FIRST: decision rights.
WHO MUST AGREE: the payments department head.Notice what the test leaves out. It doesn't tell you how to fund the work, and it doesn't ask whether the events are any good. It asks whether the structure moved, and it points at the one conversation that would move it.
Try this next week
Pick one unit you'd call scaled and run the Silo Test on it with two or three people who were there for the rollout. Sort by what moved, not by what was announced. Then take the one structural answer you'd change first to whoever would have to agree, and ask them what actually changed when the unit scaled.
If you want help working through the structure side, that's the organizational design work we do in agile coaching. When I start a transformation now, I usually start there, before anyone renames a team.
If your Silo Test comes back all mechanics and you're not sure where to start, send me a note if you need help.