Steering committees never lack a plan.
They lack the decision.

Why action programmes rarely fail in execution – and almost always because of a decision that nobody takes on.

Every institution under pressure has a list of measures. It is well maintained, it has traffic-light colours, it has named owners, and it is on the table at every meeting. What it lacks is impact.

That is no criticism of the people working on it. Most of them are working hard. They are just working on points that someone else should have decided long ago.

That is why, in a new mandate, I do not first look at how many measures are open. I look at how many of them are waiting for a decision that the person responsible is not allowed to take. That share says more about a programme than any progress figure.

The steering committee is the heart of any project governance. If it is only reported to, rather than deciding, the programme has governance on paper and none in practice.

The list is not created at the whiteboard

In a bank, measures come from outside. From findings of internal audit. From the auditor’s management letter. From a supervisory examination. Plus the points from the institution’s own initiative.

This means that a measure there is not a good idea, but a documented commitment with a deadline. That has two consequences you need to know before talking about prioritisation.

First: the list grows from outside. It is not the result of prioritisation, but a collection of commitments from different sources that were never weighed against each other.

Second: nobody deletes anything. Removing a line looks like non-fulfilment. So it stays – even when the reason for it was dealt with a year ago.

In this environment, prioritising therefore does not mean leaving things out. It means setting the order, allocating capacity and justifying why number forty will only be tackled in the second quarter. That justification holds up if it is written down. Unwritten, it is an excuse.

Three places where it gets stuck

The person responsible is not allowed to decide. There is a name on the list. What the list does not say: the measure touches a responsibility on which a committee decides. The person whose name is there can push, remind and prepare – but cannot decide. They carry responsibility for a result they are not allowed to bring about. Every measure therefore needs a second sentence: what this person may decide alone, and above which threshold it has to go to the committee.

The deadline does not know the meeting calendar. The deadline is 31 March. The committee meets on 9 April, and the papers must be submitted ten working days beforehand. The measure was overdue on the day it was planned – and nobody noticed, because deadlines are thought through forwards instead of backwards. I calculate deadlines back from the meeting calendar. Then it is immediately clear which commitment could never have been met in the first place.

“Implemented” is confused with “working”. The work instruction has been changed, the training has taken place, the box has been ticked. Whether the process actually runs differently in day-to-day business, nobody has checked. For me, a measure is only done when there is evidence from operations and not from the project: a sample, a key figure, a review after ninety days.

What I change in the first two weeks

Four interventions that cost little and quickly become visible:

I keep a second list. Not the measures, but the open decisions: who takes them, in which committee, by when, and what is blocked as long as they are outstanding. This list is usually short – and that is exactly the uncomfortable news. Five or six open decisions often hold up half the action plan.

I take the traffic lights out of the status report. Colours invite negotiation. Instead, four values: running, waiting for decision, overdue, completed. “Waiting for decision” is the most important of them, because it hands responsibility back to where it belongs.

I allocate capacity, not priorities. Anyone who prioritises without taking something away from someone has merely sorted. The measures that carry the result need named people and cleared time; all the others get an honest, later date.

I give every blockage an expiry date. If a decision is still outstanding after two meetings, the item goes onto the board agenda without further discussion. That is not a threat; it is an agreement you make at the start, while nobody is yet affected.

I have described elsewhere how this plays out in a deadlocked case: why half a position can paralyse an organisation.

What AI takes off your hands – and what it does not

Quite a lot, and reliably: matching minutes against the list of measures, finding duplicates and dependencies, making overdue items and contradictory information visible, condensing reports. In a large programme, that saves real working days.

What it does not take off your hands is the part where things fail: stopping a measure, replacing a person responsible, pushing a priority through against resistance. That takes someone who answers for the consequences. More on this in AI writes the plan in four minutes.

Two questions for your own list

Take your current action plan and answer two questions. You do not need anyone from outside for this.

First: how many measures are waiting for a decision that does not lie with the person named as responsible?

Second: how many deadlines fall before the next meeting of the committee that has to approve them?

If you cannot name both figures off the top of your head, that in itself is the result. And it is a good result, because it can be changed within a week.

→ To the project check: twenty statements, four minutes, no registration

And if you would rather talk about it straight away: thirty minutes, without obligation. At the end, you will know where you stand.