There is a peculiar kind of corporate self-deception that goes by many names: digital transformation, AI readiness, data-driven culture. The vocabulary shifts every few years but the underlying move is always the same. A company spends an extraordinary amount of money building analytical infrastructure on top of a foundation it has never actually bothered to inspect.
The ERP system is that foundation. And in most organisations, it is a mess that nobody wants to look at directly.
Take a purchase order worth two million dollars sitting in an approval workflow for eleven days. The system records the delay with perfect fidelity. What it does not record is why. Nobody asked the approver to explain the hold. No reason code was triggered. The system simply noted the elapsed time and moved on, and that explanation evaporated into the ordinary noise of operational life. Eleven days, two million dollars, no story.
That is not a software failure. The ERP system did exactly what it was configured to do. It is a governance failure; which is a polite way of saying it is a discipline failure, which is an even politer way of saying nobody actually cared enough to ask.
Every ERP environment generates hundreds of signals like this, continuously. Goods receipts posted while warehouse putaway confirmations never arrive. Obsolete materials still generating planning proposals months after anyone last sold them. Purchase orders raised without valid contracts behind them. Sales orders sitting motionless for ten, twelve, fifteen days with nothing moving in the logistics pipeline. Manual journal entries sailing past approval thresholds. Suspense balances accumulating quietly in the general ledger like sediment. Production equipment going down with nobody recording the reason inside the system that is supposedly running the factory.
None of these are exotic analytical problems. They are basic operational questions that the system should have been asking all along: why is this happening, and what does it mean? A structured reason code at the moment of deviation costs almost nothing to implement. It creates something invaluable: a traceable operational narrative. Twenty minutes of configuration, a lifetime of accountability.
Instead, organisations skip this entirely and go straight to the fashionable part. Data lakes. Process mining. Predictive planning. AI programmes with slide decks full of architecture diagrams and promises about machine learning transforming operational insight. These initiatives are not wrong, exactly. They are simply being bolted onto a system that has never learned to explain itself.
Artificial intelligence cannot interpret causes that were never recorded. Process mining cannot reconstruct explanations that nobody bothered to capture. You can build the most sophisticated analytical stack in the industry and it will faithfully, expensively, tell you that your approvals are slow and your warehouse confirmations are missing. Which you already knew. Which you have always known.
The actual insight begins much earlier. It begins the moment someone in the workflow is asked: why did this happen? That question, asked systematically, through reason codes that feed into standard reports, transforms an ERP system from a passive transaction ledger into something far more interesting. It becomes a mirror. And the reflection it shows is not the idealised process that lives in documentation; it is the organisation as it actually behaves.
The CEO believes the order-to-ship cycle averages five days. ERP data may show it ranges from four days to thirty-nine, depending on circumstances nobody has ever formally explained. That range is the story. Reason codes make the story legible: approval bottlenecks at month-end, warehouse confirmation delays in a specific facility, commercial negotiations holding up fulfilment, incomplete order data forcing manual intervention. Suddenly the variability has causes, the causes have owners, and the owners can be spoken to.
Building this capability does not require a programme, a budget, a steering committee, or a vendor. It requires optimised queries and analytical views that read directly from live ERP transactions and prompt users for explanations when deviations occur. No separate data warehouse. No pipeline to maintain. The analysis runs against the system that already exists, because the information was already there. It just needed someone to ask for it.
Once organisations start asking, something predictable happens: accountability becomes visible. Approval delays acquire traceable causes. Patterns emerge. Why do certain approvals stall at month-end? Why does one warehouse consistently lag on goods receipt confirmations? Why do particular sales orders sit dormant for weeks? These questions, which once dissolved into the background noise of operations, now have answers. And answers, once visible, are very difficult to continue ignoring.
Operational discipline improves. Not because anyone delivered a motivational speech about process excellence, but because the system is now holding up a mirror and people can see themselves in it.
There is a certain irony to the whole thing. Organisations spend millions on analytical platforms that promise to reveal how they operate, while the most powerful governance instrument they own sits inside the ERP system they already paid for, waiting to be used properly. A handful of reason codes. A few well-written queries. An insistence that deviations must be explained before the workflow moves forward.
That is the billion-dollar idea. It just does not require a billion dollars.
