Ask a leadership team why the last transformation did not stick and you will get one of two answers. The people did not adopt it, or the process got in the way. Both are usually sincere. At least one of them is usually wrong.
The uncomfortable part is that every team believes it already knows which. Roughly half of them are wrong, and the ones who are wrong are wrong with total conviction. Seniority does not help. Nor does being close to it. If anything, being inside it is the problem, because both conditions feel identical from in there.
So we stopped arguing about it and started asking the same seven questions at the start of every engagement. They take about fifteen minutes. They do not diagnose anything, and I will come back to that at the end. What they do is make the right argument start quickly, instead of the wrong one running for a quarter.
Here they are, and what each one is actually listening for.
1. When something does not land, what gets blamed first?
Not what is true. What gets said in the room.
The answer tells you what the leadership already believes, which matters because the next six questions will either confirm it or quietly take it apart. If the room blames the process, you are usually sitting with people who have run out of patience with their own structure. If it blames the people, you are usually sitting with people who built something sound and cannot understand why nobody is using it.
The most revealing answer is the fourth one: nothing gets blamed. It just stops being mentioned. That is a business where things fade rather than fail, and nothing marks the moment.
2. Your last significant platform or tool rollout. Is it used the way it was designed?
Designed, not licensed.
This is the cleanest single tell there is. An organisation can be very good at acquiring capability. It can buy the tool, deploy the platform, document the process and produce a slide showing all three. What it cannot do is change what people actually do at eleven o’clock on a Tuesday.
Roughly 95% of enterprise AI pilots produce no measurable impact on profit, according to MIT Project NANDA. The technology mostly worked. The behaviour never changed. If the honest answer here is “bought, deployed, barely used,” you already know a lot.
The opposite answer is just as useful. “We have not rolled one out, we work on relationships” is not a confession. It is a different problem entirely.
3. Can the people furthest from power act without asking permission?
This one came out of work we did for EY in 2019, on whether a city can be designed to be iconic on purpose. The test that emerged was that a place works when women, girls and children feel free simply to be. Pointed at an organisation, it gets blunter.
If they cannot act, the structure has crowded out the people, whatever the values statement says. If they can act but nothing they do accumulates anywhere, you have the opposite problem and a lot of wasted goodwill.
There is a third answer we hear often. In theory yes, in practice nobody tries twice. That is an organisation that has taught its own people what happens when they try.
4. Take your three strongest initiatives from last year. Did the number move?
The headline number. Not the project metrics.
“Each one worked. The total did not move” is the signature of an organisation with conviction and no value chain. We modelled a national foundation whose board was debating three strategies, all of them credible, all of them with advocates in the room. Under existing conditions all three produced 0.0% measurable impact. Not because the strategies were bad. Because there was not one complete chain in the organisation running from a programme through to an outcome anyone would fund.
Choosing a better programme changes nothing when the conversion structure does not exist.
5. Who has the highest customer satisfaction and the smallest budget?
If a name comes to mind immediately, that is the answer. If you would have to go and look, that is also the answer, and it is the more interesting one.
At Primo Waters, a business that had come through a run of acquisitions, the delivery drivers held the highest Net Promoter Scores in the entire company and were being managed as a cost line. The people customers trusted most had the least strategic role in the business.
Nobody had hidden that. It was in the data. Nobody had looked, because there was no reason to go looking in the place where the trust turned out to be.
6. Could you state today what the thing you want to improve is currently worth?
A number, not a direction of travel.
This is the least glamorous question on the list and it eliminates more wasted work than the rest put together. Without a baseline, “the team found it useful” is the only finding available at the end, and it is not a finding.
Most organisations cannot answer it. The common version is “we have the data, nobody has ever pulled it together,” which is a structural gap rather than a people one. It is also usually a week of work rather than a project, and it changes what every subsequent conversation is worth.
7. Does a new way of working need permission from the same people who were the bottleneck before?
This one decides more than it looks.
If a new process routes through the committee that was already the constraint, the structure did not change. A step was added to it. The organisation will experience this as the change failing, and it will usually blame the people who were asked to adopt it.
What to do with the answers
Two patterns come out of this, and they need opposite responses.
The answers lean one way. Good. That is your hypothesis, and it is worth about as much as a hypothesis should be. One side of the line means the structure exists but nobody uses it as designed, and the work is finding where the trust actually sits. The other side means there is real conviction and nothing compounds, and the work is building a chain that turns activity into evidence and evidence into funding.
The answers do not lean at all. This is more common than you would think, and it is not a failure of the questions. It almost never means the organisation is balanced. It usually means different parts of it are in different places and we have been talking about the business as though it were one thing. The next move is to pick the division, region or product line where the problem is sharpest and run the same seven questions against that alone.
Why this does not settle anything
Here is the part that is inconvenient for us as much as anyone.
Seven questions cannot diagnose an organisation. They are answered by one person, usually the person who called the meeting, and one executive’s read is not the business. The honest thing to do with two senior people is run it separately with each of them, because where they disagree is a better conversation than where they agree.
And even a clear answer is a guess. You cannot argue an organisation into knowing its own blend. You have to show it, at contained scale, with real data, in a way that is cheap to be wrong about. That is what a pilot is for, and it is why the output of fifteen minutes should be a hypothesis and a test, never a verdict.
What the questions buy you is speed. They do not tell you the answer. They tell you which argument is worth having, which is most of what is lost in the first three months of a transformation programme.
We ask them on a shared screen in the first half of a thirty minute call. You are welcome to ask them yourself.
Fourth in a series of short essays answering the questions that come up most often in client work. The others asked whether your problem is structure or people, why most AI pilots fail, and what happens when you leave.