The most valuable decision in a transformation is what to stop

Ask a leadership team how many change programmes are running and the answer is usually three, perhaps four. Count them properly — every initiative with a sponsor, a budget or a recurring meeting — and the number is rarely smaller, while the same names turn up on most of them.

That overlap, not the technology, is what decides whether an AI programme lands.

Why adding is easy and stopping is not

Starting costs nothing visible. A new initiative needs a sponsor and a kick-off. The cost — hours taken from people already working on something else — appears months later, spread across other programmes as delay, and is never attributed to the decision that caused it.

Every running programme has a defender. Somebody sponsored it, somebody’s objectives depend on it, and stopping it reads as an admission. So programmes are rarely stopped; they are slowed, which keeps all of the overhead and none of the progress.

Capacity is measured by asking. Ask people how busy they are and the answer is always the same. A plan built on that answer simply assumes the capacity it needs.

What a capacity assessment actually counts

Not a survey. Four things, measured.

Every programme running. Including the ones nobody calls a programme: the system migration, the reorganisation, the audit remediation. They draw on the same people.

Who is on more than one. The names that appear three or four times are the real constraint of the organisation. Usually a small group across operations, finance and IT, and usually the same people the AI programme needs most.

What each one is waiting for. Programmes stalled on a decision or on a person consume attention without producing anything.

How much change the organisation can take this quarter. Stated as a number of changes, not as a percentage of anyone’s time. Two done properly is a reasonable answer for many organisations; five started is a common one.

The result tends to surprise the people who commissioned the programmes. Nobody hid anything. Each programme was approved on its own, against its own business case, and nobody ever added them up. Laid side by side, they show the same person in finance with four weekly steering meetings and no free afternoon.

Infographic

From a list of programmes to a decision

1. 01Count everythingEvery initiative with a sponsor, a budget or a recurring meeting.
2. 02Map the peopleWho is on more than one programme, and which ones.
3. 03State capacityHow many changes we can take this quarter, counted in changes.
4. 04Decide what stopsWhat goes, what waits on a condition, what ends.
5. 05Say it plainlyTo the people whose work stops, with the reason.

How to stop something without losing the people

Stopping is a decision, not an announcement. Handled well, it is the moment the organisation starts taking the programme seriously.

Defer with a condition, not with a vague later. “This waits until the first change is live and finance is through the year-end close” is a decision. “Later” is an invitation to keep working on it quietly.

Say what happens to the work done so far. Documented, archived and owned by somebody, so that restarting does not mean starting again.

Release the people visibly. If the stopped programme’s team is quietly reassigned to the next thing, nothing has been freed. The hours have to show up on the programme that needs them.

Let the sponsor of the stopped programme deliver the message. It is harder, and it turns a loss into a decision they made rather than one made to them.

The meeting that decides the outcome

In most transformations there is one meeting where the order is agreed and, more importantly, where something stops. Everything after it is execution. If that meeting ends with every programme still running at a slower pace, the company will get a little of everything, late — the outcome nobody chose and everybody receives.

The people in that meeting matter more than the agenda. It needs whoever can actually stop something: the sponsors of the competing programmes and the person they report to. Without them, the meeting produces a recommendation to stop, which is not the same thing.

A useful test for the sponsor: name the programme you stopped to make room for this one. If there is no answer, the AI programme is being funded out of everybody’s evenings.

Where we start

In an AI transformation engagement the capacity assessment comes first, measured rather than asked, and it ends in a sequenced portfolio: what goes first, what is deliberately deferred and what stops. It is usually the least comfortable document in the programme, and the one that decides whether the rest of it works.