In an earlier piece we said there was a second problem sitting underneath the first one, and that we would come back to it. This is that piece.
Picture the kickoff for a new data and AI strategy exercise. Fresh energy, a workshop schedule, someone walking the room through the plan. And about twenty minutes in, something nags at you. You have been here before. Not somewhere like here. Here. Three years ago, different faces, the same questions, the same whiteboard, the same headings that will end up on the same slides. Somewhere in the building, or in an inbox, or on a shared drive nobody remembers the shape of, there is a document that already answered most of this. You just cannot lay your hands on it. Or the person who wrote it has left.
The loop
Once you see this, you cannot unsee it. Organisations do not fail to think about data and AI. They think about it repeatedly, and they lose the thinking every single time. Discovery, diagnosis, prioritisation, a roadmap. Done properly. Presented well. Approved. And then, quietly, gone. Eighteen months or three years later the appetite returns, the budget is found, and the whole cycle runs again from close to zero. Most companies are not under-thinking their data and AI strategy. They are re-thinking it, over and over, and paying full price each time.
Where the thinking leaks out
There are two leaks.
The first is the deck. A strategy gets turned into a sixty-slide artefact. It is presented to the board, it is admired, and then it is filed. A deck is where thinking goes to be congratulated and then forgotten. Part of the reason nobody reopens it is that a slide is a terrible container for reasoning: it holds the conclusion but not the argument, the what but not the why. Read it a year later and you can see what was decided, but not what it would take to change your mind, which is the only part that actually helps you next time.
The second leak is people. The real intelligence in any strategy is not the recommendation. It is everything around it: the reason this option was chosen over that one, the constraint that killed the obvious idea, the trade-off that was weighed and the judgement that settled it. That lives in heads. In the head of the sharp consultant who ran the engagement, and in the head of the employee who quietly held it all together. When the engagement ends, the consultant leaves and takes it with them. When the employee moves on, so does the map. What you keep is the deck, which, as we just established, gives you the answer but not the thinking.
Why data and AI is the worst case
Here is the trap. The comforting story is that this thinking has to be redone because the field moves so fast that any strategy is stale within the year. There is a sliver of truth in that. The tooling does move quickly. But the foundational work does not. What data you actually hold. Which decisions in your business genuinely matter. What your real constraints are, in systems, skills and appetite. What you decided last time, and why. That layer moves far more slowly than people assume, and it is the expensive part to rebuild. So when the loop runs again, you are not mostly redoing work because the world changed. You are re-discovering answers that were still true. The waste is not the pace of AI. It is throwing away valid conclusions and paying to reach them a second time.
The honest part about consultants
We want to be careful here, because it would be easy to read this as a swipe at consultants, and it is not. Good consultants do genuinely valuable work, and we say that as people who do it. The problem is not the people. It is structural, and it is worth being honest about, because the structure is the whole point.
Consulting is built to deliver insight, not memory. The deliverable is a recommendation, a direction, a decision. It is not, and was never designed to be, a system that holds that thinking inside your organisation once the team has gone. So the value is real and the evaporation is built in, at the same time. It is not a failure of the engagement. It is what the arrangement produces: sharp thinking, delivered, and then no mechanism for keeping it. Blaming advisers for that is like blaming a chef because the meal does not last the week. That was never the deal. The gap is not bad advice. It is that nobody owns what happens to the advice afterwards.
The cost nobody counts
The repeated spend is the visible cost. You can see it, it has invoices attached, and it is the one people complain about. It is also the smaller half.
The larger cost is invisible, and it does not show up on any line. It is the decisions taken without the context that had already been established and then lost. It is the initiative restarted from scratch because the reasoning behind the last version went home in someone's head. It is an organisation that never compounds its own understanding of itself. That last one should genuinely bother you, because compounding is the whole game. Your brand compounds. Your customer relationships compound. Your intellectual property compounds. The one strategic asset most firms actively allow to depreciate back to zero, again and again, is their own accumulated thinking about data and AI. You would never let your brand equity reset every eighteen months. You do exactly that with this.
Think about it once, and keep it
The fix is not better consultants, and it is not better decks. You can hire the sharpest firm in the market and buy the most beautiful slides ever produced, and you will still be back in this room in three years, because neither of those things is built to make the thinking stay.
The fix is to stop treating the thinking as a project and start treating it as an asset: something the organisation owns, holds, and adds to, so that the next time the question comes round you begin from what you already know rather than from an empty room. The goal is not to think about your data and AI strategy one more time, only better. It is to think about it once, and keep it.
Stop renting thinking that leaves. Start owning thinking that stays.