What Is Decision Latency?

Decision latency is the total time between a threat's first signal and a completed organizational response. Not the moment of choice — the whole chain around it: seeing the signal, naming the threat, committing the call, executing it, and pivoting when the situation changes. It is the single most predictable cause of preventable loss I saw in 17 years and 16,000 critical calls of frontline command — and the one number almost no organization has ever measured.

The definition, precisely

In the V=A/L framework — Command Velocity = Accuracy ÷ Latency — accuracy is whether the call is right, and latency is how long the organization takes to land it. The definition has a sharp edge: latency starts at the first signal, not at the moment leadership becomes aware. The clock was running while the anomaly sat unread, while the alarm was normalized, while the email waited for Monday. Organizations that measure from "when we found out" are grading themselves on the shortest, most flattering segment of the timeline.

The five components — where the seconds actually die

• Recognition lag — the threat exists; nobody has named it. East Palestine's failing wheel bearing crossed three trackside detectors over thirty miles, reading 38°, 103°, then 253° above ambient. Only the last crossed the alarm threshold. The signal existed the entire time; the threshold design dismissed it.
• Assessment lag — it's detected, but misread. Lahaina's morning fire was declared 100% contained; by 3 PM it was the deadliest US wildfire in a century. A wrong classification holds longer than no classification, because it switches the searching off.
• Decision lag — understood, but uncommitted. Maui had 80 outdoor warning sirens that day. None were activated. The infrastructure worked; the authority to use it stalled. 102 people died.
• Execution lag — committed, but dropped in the handoff. When the M/V Dali lost power in Baltimore, police stopped bridge traffic within about 90 seconds of the mayday and saved every motorist. No channel existed to warn the road crew on the span; six men died. Same event, same minutes — one executed handoff, one missing one.
• Adaptation lag — the situation changed; the plan didn't. The 2024 CrowdStrike outage hit every airline identically. Most restored in a day; one took five, at a reported cost near $550 million. The shock was equal. The re-planning latency wasn't.

The compounding rule matters: early latency is leveraged latency. A recognition stall delays every stage after it, which is why two organizations with the same total headcount and the same plan can have wildly different outcomes from the same event.

Latency vs. the threat's timeline

The pass/fail line isn't a benchmark from a consultant's slide — it's physics. Every threat has a timeline: how long the chemistry, the fire, the algorithm, or the water gives you. Knight Capital's runaway code made roughly 1,500 erroneous trades per second; the humans needed about 15 minutes to recognize and kill it. $440 million gone in 45 minutes. The firm's latency was ordinary. The threat's timeline was not — and the threat doesn't grade on a curve.

How to measure yours

One real scenario, the people who'd actually respond, and an honest clock on each of the five stages. Sum the delays. Compare against the threat's timeline. If you want the structured version: the free V=A/L Latency Audit walks you through ten questions and returns your Velocity Score, grade, and dominant bottleneck in about ten minutes — no signup. The half-day Velocity Diagnostic does it live with your leadership team, against your own scenario, with a documented gap report and a 90-day roadmap.