AI in management is transforming how businesses analyse data and explain complex events. But there is a problem: explanation is not the same as understanding. AI can generate insights instantly. What it cannot replace is structural thinking in management.
Without it, even the most advanced tools fail to produce real impact.
Nothing in corporate life is easier to produce than an explanation after the event. We now have AI to help us with that noble tradition.
So, why does management still need structural thinking? Because dashboards, forecasts and AI tools mostly describe flows, while business performance is often governed by stocks, constraints, feedback loops and delays.
AI is becoming very good at explaining what just happened in a business. I am not sure that is the same thing as helping management understand what is going on.
I have spent enough years around financial reporting, dashboards and financial models to know how comforting a good explanation can be. A bad month gets blamed on pricing, competition, execution, seasonality, slower collections, weaker conversion, or some mix of all six. The explanation may be sensible. It may even be correct. But I have seen sensible explanations coexist quite happily with bad decisions.
System structure is the source of system behavior.
Donella Meadows, Thinking in Systems
That is because businesses do not run on explanations. They run on structure. Revenue, margin, cash and churn are what the system reveals at a given moment. They are visible movements, not the system itself. The deeper forces usually sit elsewhere: in stocks that are being built or depleted, in delays between cause and effect, in constraints that tighten quietly, and in feedback loops that managers notice only when they become painful.
By structure I do not mean anything mystical. I mean the mechanisms that make one business behave differently from another even when the surface numbers look similar.
Some of those mechanisms are stocks: cash, inventory, trust, know-how, backlog, receivables, organizational attention.
Some are constraints: production capacity, management bandwidth, financing limits, hiring speed.
Some are feedback loops: lower service quality creates more complaints, which absorbs more time, which lowers service quality further.
And some are delays: the hiring decision shows up in productivity months later; looser credit control shows up in cash pressure only after revenue has already looked strong.
I do not mean this in some mystical, consultant-flavored sense of “the system.” I mean something very ordinary. A company can grow and become weaker at the same time. I have seen that in finance work more than once. Sales rise, everyone relaxes, and then cash starts to feel tight. The dashboard explains the variance. The AI summary explains it better. But the real issue is structural: receivables are stretching, working capital is being consumed, and growth is outrunning the company’s ability to finance itself. This is where structured analysis and decision support become critical mnot just to explain what happened, but to change what happens next.
This is one reason management so often misreads performance. It focuses on events because events are visible, recent and easy to discuss. A KPI turns red. A team misses target. Margin drops two points. A forecast is revised. Much of corporate life is built around reacting to these things. Power BI makes them visible. Power Platform can route them, alert them and escalate them. AI can now summarize them in seconds with admirable fluency. None of that is trivial. But none of it guarantees that management is looking at the right level of reality.
Here’s a snapshot of Donella Meadows’ thinking that I find particularly insightful:
It’s endlessly engrossing to take in the world as a series of events, and constantly surprising, because that way of seeing the world has almost no predictive or explanatory value.
Donella Meadows, Thinking in Systems
The deeper problem appears when the business stops behaving in straight lines. Managers like linearity because it feels fair. A little more effort should bring a little more result. Twice the investment should bring something close to twice the return. But many business systems do not behave like that. A little more growth can create a lot more pressure on cash. A little more process can create a lot more friction. A little more marketing can lift demand, until it starts to damage trust or flood an operation that was already close to its limit.
That is why nonlinearity matters. Not just because outcomes become disproportionate, but because systems can change mode. What looked like healthy growth becomes congestion. What looked like operating discipline becomes bureaucracy. What looked like financial momentum becomes working-capital stress. The manager who thinks only in increments sees movement but misses the shift in regime.
And that, I think, is where AI will make management both more capable and more vulnerable. It will explain events beautifully. It may even explain them better than most analysts. But management still needs something harder than a good explanation of the visible. It needs a grip on the structure underneath.
AI may explain the visible movements with extraordinary fluency. Management still must see the structure that produced them. Otherwise, we’ll become better at describing performance than at changing it.
This is why I rarely start with the dashboard. I like to start each project by understanding what the business is actually doing: where stocks are being quietly depleted, where feedback loops are building pressure that KPIs don’t yet show, where the delay between decision and outcome is misleading management. Only after that does it make sense to build anything that measures or visualizes. If you recognize this problem in your business and want to explore what a structural analysis would look like, let’s talk.
