The Road Less Traveled: Why the Best Calls Start Where the Data Stops

Almost every conversation about global infrastructure this year runs along the same lines: ports, grids, subsea cables and data corridors. The physical and digital rewiring of how the world moves goods, energy and information is a real story, and a big one. What it tends to leave out is the part that decides whether any of it holds up under pressure: the judgment of the people running it.

It happens again and again. Infrastructure gets approved, funded and built. The business case is sound and the data behind it is solid. Then two organizations build from the same blueprint, read the same dashboards and end up in very different places. What separates them is rarely the evidence; it’s what they do with it.

Organizations stress-test almost everything. They load-test networks, pressure-test systems, stress-test supply chains, reroute fleets, rehearse what happens when a port closes or a supplier fails. They model buildings, audit processes and war-game their cyber defenses. Anything you can gather data on, you can test, and most organizations test it well. Artificial intelligence (AI) is simply the newest system on that list, and it’s being tested hardest of all: only 18% of corporate affairs teams feel prepared to manage a deepfake or AI-driven misinformation incident, while 43% say they’re not very prepared. The one layer no one runs a stress test on is the judgment of the people who have to make the call when the data runs out.

The more organizations optimize systems and processes, the more valuable human judgment becomes. When every organization draws on the same models and the same data, their outputs start to converge. Similar conclusions, similar plans, similar moves. Test everything on the same inputs and you get very good at reaching the same conclusions as everyone else.

By now the standard response to this is familiar: pair the machine with human judgment. It’s become the default line in nearly everything written on AI this year, and on its own it doesn’t take you far. The harder question, and the one fewer organizations can answer, is whether they’ve actually embedded that judgment into how they decide, or whether they just reach for it once the data runs out.

This is the road less traveled. Building the foundation on what’s known and what the data is showing; that part isn’t optional. But the organizations with the edge are the ones willing to layer human judgment and intuition on top of the evidence rather than surrendering the decision to it. The data tells you where the crowd is going, judgment is how you decide whether to follow.

APCO’s own Global Corporate & Government Affairs Collaborative captured the mood plainly this year: the function has never been more strategically important, and there’s no reliable playbook for anything anymore. No playbook means no substitute for judgment. You can automate inputs, but you can’t automate the decision about what they mean and what to do next.

This is where foresight comes in, and where most organizations have a blind spot. They stress-test their physical and cyber infrastructure as a matter of routine, but not many stress-test their decision-making. They run scenarios for what might damage the building, but not for how they would make decisions under pressure, with incomplete evidence and the clock running. Good foresight doesn’t try to predict which scenario will come true, rather its value is in rehearsing those calls before you’re forced to make them for real: where does data stop and judgment start, who’s trusted to decide when the evidence runs out and how fast can they do it? An organization that has practiced those calls in advance moves very differently from one meeting them cold.

So, what is a test worth running? You’ve mapped what you own and what could go wrong with it, but have you mapped how you decide? A resilience audit for judgment, not just for assets, is the harder piece of work, and it’s the one far fewer are willing to do.

The world keeps getting better at connecting places and standardizing what everyone knows. The question worth sitting with is simpler: when the data runs out and the call is yours, will you have honed the judgment to make it?