How ungoverned is your acceleration?

Nine questions about how work gets built, chosen and seen in your company, scored on two axes – how fast you are already moving, and how much of it you can govern. The result is a position, in your browser only, and a suggestion for what to read next.

The nine questions

When someone in the business has a clear idea for a report, a tool or an interface, how long until they can use it?

• Nobody could say; it depends on who is asking
• A quarter or more, once it earns a place on a roadmap
• A few weeks, if it fits into a sprint
• Days, and often the same day

Who in your company can build a working thing without going through engineering?

• Building is not what our people do
• Only engineering, through a request
• A few people, quietly, in tools nobody else can see
• Many people, and they do it where others can build on it

What do your most senior engineers spend most of their time doing?

• Reviewing plans, estimates and requirements
• Writing and reviewing code by hand
• A mix – some direct agents, most still write
• Directing agents and judging what they build

How does work get chosen?

• Steering committees and gates at every step
• An annual budget and a roadmap
• Quarterly planning, with some room to move
• Teams pull from a queue of outcomes they own

Who owns the customer – or the product, or the order – end to end?

• Nobody could say
• Several functions share it, each with a piece
• One function, mostly, without the data or the interfaces
• One named owner: the thing, its data and the interface everyone else uses

How many projects have been "almost done" for more than a year?

• We stopped counting
• Several, and everyone knows which
• One or two
• None – stalled work is ended or given an owner

How many AI agents or AI-built tools are running in your company right now, and who could tell you?

• We have asked, and got different answers
• No idea
• Roughly – someone could find out in a week
• We keep a registry, with an owner and a cost for each

When a department builds or buys its own AI tool, what does it inherit from the company?

• Nothing, and we find out later
• Whatever the vendor offers
• Permissions, sometimes, after a review
• Identity, permissions, logging and spend metering, automatically

Where does spend on AI, cloud and people show up together?

• Nobody has added it up
• One technology line, once a year
• By department, after the fact
• Per team and per outcome, visible daily

The four positions

Rationing

Slow and ungoverned. The old machinery still holds, and the change will start underneath you.

The machinery that rations the capacity to build is still holding, and from the executive floor that can look like control. It is not. The same tools are on every laptop in the company, and the building will start underneath you, department by department, before anyone decides anything – I have watched it begin in the quarter after a board said "not yet." The advantage of being here is that you can still choose the order: governance first, then the first domain, then speed, which is the order the model was built in.

Leaking

Fast and ungoverned. People are building everywhere and nobody can see it.

Speed has arrived and it is arriving badly: people are building, agents are running, and the company cannot say what they are doing, what they are spending or what data they reach. This is where most companies I talk to sit, and it is the position the diagnostic engagement is built for, because the stalled and duplicated portfolio is largest here and the fix – a control plane and one owned domain – is roughly the same size at every company. Read the model, then let us find your number.

Held back

Slow and governed. You have the controls, and you still ration who gets to build.

You can see and govern what gets built – identity, spend, ownership – and you still ration who gets to build it. That is the rarer position, and the easier one, because the hard part is already in place. What stands in the way is the machinery: the roadmap, the planning cycle, the line between the people who want a thing and the people who build it. Removing it is what the four shifts describe, and one domain proves it in a quarter or two.

Governing the change

Fast and governed. Few companies are here, and I would test the answer.

Few companies are here, and a self-assessment is generous, so I would test it: take the last thing the business built for itself and trace who owns it, what it cost and who else builds on it. If all three are visible in one place, the conversation is not about the first domain, it is about the next one, and about the platform capabilities that do not yet exist anywhere. That is the last rung of the offer, and the part of this work I find most interesting.

A self-assessment is a position, not a number. It tells you which of four situations your company is most likely in, based on how you answered nine questions about it, and it will be generous in places where you were generous with yourself. The number – what your stalled and duplicated portfolio is actually worth, and which project is the smoking gun – only comes from looking at the portfolio itself, which is what the diagnostic engagement does in a few weeks.

Two things I would say about the axes. Speed measures how fast the company is already moving, not how fast it could, because the tools that make building abundant are on every laptop already, and the honest question is whether people are using them. Governance measures how much of that movement the company can see, own and pay for in one place – identity, ownership, spend – which is the part that turns a leak into an operating model. The dangerous quadrant is fast and ungoverned, and it is also the most common one, because acceleration arrives on its own and governance has to be chosen.

Nothing you answer here leaves your browser. If you would like to talk about where you landed, the button below opens an email with your position in the subject line and nothing else.