A regional utility's grid balancing system watches load across six hundred substations in real time, shifting capacity automatically the moment a transformer runs hot or a feeder nears its rated limit. For four years, the quarterly operations report handed to the board carried one recurring line. Grid AI performance: nominal. In that same four years, the system executed more than two thousand automatic reallocations, three of them in direct response to conditions that would have caused a localized outage within minutes had the system not intervened first. None of the three ever reached a board packet. Nobody buried them. Nominal was simply the only word anyone had ever been asked to produce.
The engineers who watched the dashboard could describe all three incidents from memory, each with a timestamp, a threshold crossed, and the minutes between detection and correction. That detail had never been difficult to produce. It had simply never been requested. The board received a status. The organization held the facts underneath the status, unasked, the way most organizations do.
What changed this pattern at that utility was not a new sensor or a new report format. It was a single audit committee chair who stopped asking whether the grid AI was operating normally and started asking a different question entirely: what would not normal look like, in specific terms, and how quickly would the organization know it was looking at one. The first time she asked, management produced a vague answer about thresholds and alerts. She did not accept it. She kept asking, meeting after meeting, until the organization could name the exact conditions that would count as a warning and the exact minutes it would take to recognize one.
This is the distinction most boards never draw, and naming it plainly is worth the discomfort it causes. A board that asks for a status report is managing an existing relationship with information. A board that insists on defining, in advance, what would count as evidence of trouble is doing something else entirely. It is raising the organization's own standard for what oversight requires. This is the Expectation Elevation Model at work in its most practical form. The highest function of a board member is not to grade the report the organization chose to produce. It is to help the organization see, often for the first time, what a genuinely rigorous account of itself would actually contain.
The reason nominal survives as long as it does is structural, not a failure of character on either side of the table. A board meets on a calendar. Management prepares materials for that calendar. Neither party has an incentive to volunteer detail nobody asked for, and the sentence least likely to generate a follow up question is always the safest one to write. Three incidents, each with a timestamp and a dollar figure attached to the outage it prevented, invites scrutiny that nominal never will. Left alone, every organization drifts toward the sentence that ends the conversation. The only correction is a standing, specific question, asked on a fixed cadence, until the sharper answer becomes routine rather than a surprise.
What the utility discovered, once the chair raised the bar, is that the monitoring had been adequate all along. The gap had never been in the system watching the grid. It had been in what the board had ever required the system to tell it. Once the question changed, the same technology that had quietly prevented three outages became the clearest evidence the company had that its oversight was real, because for the first time someone outside the control room had actually asked to see it.
This is uncomfortable work in a specific way most governance advice declines to name. It requires one person, in a live meeting, to keep asking a question sharp enough that a vague answer becomes visibly inadequate to everyone in the room, and to keep asking it until the organization stops treating the question as unusual. Most board members will not do this, because it risks a moment of friction nobody else at the table seems to want. The ones who do it anyway are not more suspicious of their management teams. They have simply understood that their job was never to accept the organization's own account of itself. It was to raise what the organization believed an account required.
A board does not fulfill its obligation by reading the word nominal and moving to the next agenda item. It fulfills its obligation by deciding, before anything goes wrong, exactly what would have counted as something else, and by requiring the organization to say so the moment it arrives.