You have sat in this meeting. It is the quarterly update on data and AI. The deck is good: forty slides of roadmaps, workstreams, and words like momentum and embedding. The people presenting are clearly working hard. You ask a couple of questions, you get confident answers, you approve the next round of spend. And then, somewhere on the way home, it occurs to you that you could not actually say whether any of it is working. Only that it sounded like it was.
If that is familiar, this is for you.
Why this became your problem
Five years ago it was not. Data and AI were delegated somewhere below you, and nobody asked the chief executive to account for them personally. That has changed, and quickly. Boards now ask about AI exposure directly. Investors diligence data and AI maturity at deal stage and price it into what they will pay. And the law is catching up: the EU AI Act is phasing in real obligations on the organisations that build and deploy these systems, and while its deadlines have already been pushed back, the direction has not. What was recently treated as good practice is becoming a legal duty.
The delegation hatch has quietly closed. You are now answerable for something you were never expected to understand, and pretending otherwise does not help.
Busy is not working
Here is the blunt part. Most executives cannot tell whether their data and AI strategy is working. They can tell four other things, and they mistake each of them for the same thing.
They can tell that people are busy. There is visible effort, full calendars, a team that is plainly not idle.
They can tell that money is being spent. The budget is going out of the door on schedule, which feels like progress because it looks like commitment.
They can tell that the updates sound confident. The people presenting are articulate and credible, and confidence is easy to mistake for control.
And they can tell that artefacts exist. There is a strategy. It is a big document. It took months. It must count for something.
Every one of these feels like evidence. None of them answers the only question that matters: are we making better decisions than we were, and is what we promised actually happening? Activity, spend, confidence and paperwork can all be high while the honest answer to that question is no. In fact they are often highest precisely when it is no, because effort expands to fill the space where results should be.
What being on top of it actually means
So what would it mean to genuinely be on top of this?
Not to understand the technical detail. You employ people for that, and burying yourself in it is the wrong instinct. Real oversight is narrower and harder. It is being able to answer three questions at any moment, without commissioning a two-week exercise to find out.
One. Where does the strategy actually stand, right now? Not where it stood at the last offsite. Not what the deck said in March. Where it is today.
Two. Is what we committed to actually happening? Every strategy is a set of promises. Oversight is being able to see which of them are being kept and which have quietly slipped, before someone else points it out to you.
Three. If I am asked, do I have evidence, or do I have vibes? When your board, your investor, or your regulator asks how the data and AI agenda is going, can you answer with something real, or do you turn to the person next to you and hope?
Notice that all three are about being able to ask and be answered. None of them is about doing the work yourself. Oversight is not expertise. It is visibility.
The reason it keeps slipping
There is a second problem sitting underneath this one. We will come back to it properly another time, but it is worth naming here. Even when the thinking gets done, and done well, it does not last. It goes into a deck that nobody opens again. It goes into the head of a talented consultant who leaves when the engagement ends, or an employee who moves on. Eighteen months later the same questions come round, and you find yourself paying, in money or in time, to rediscover things your organisation already knew.
So the executive is not only under-informed in the moment. They are stuck in a loop, buying the same visibility over and over and watching it walk out of the building each time.
Less detail, more visibility
When you feel this exposure, the natural response is to ask for more. More detail. Longer updates. Deeper technical briefings, so that you finally understand all of it. This is exactly the wrong move. It buries you further, and it still does not tell you whether things are working.
What you need is the opposite. Less detail, more visibility. A small number of true things you can see at any time, instead of a large pile of activity you have to take on trust.
Being on top of your data and AI agenda does not mean understanding all of it. It means being able to see whether it is working, and being able to say so out loud when someone asks. Most leaders cannot do that today. The ones who can will be the ones still standing when the questions get harder. And they are getting harder.