By James Proctor, Co-Founder and Managing Director, The Inteq Group
Decision latency is the elapsed time work spends waiting for decisions to be made: waiting for an approval, an exception review, a judgment call that needs information nobody has assembled yet. In most enterprise processes it is the dominant component of total cycle time, dwarfing the time anyone spends actually performing work. It is also, in most organizations, completely unmeasured, which is why so much process improvement effort attacks the wrong component.
What Is Decision Latency, Precisely?
Take any case that moves through your operation and split its life into two states: being worked and being stationary. The stationary state is rarely idle for mechanical reasons. The work is waiting on a decision: someone must approve, release, resolve, or judge, and until they do, nothing downstream can happen. Sum those waits across a case's life and you have its decision latency. Do it across a portfolio of cases and you will usually find the ratio uncomfortable. Work in most processes is not slow. It is stationary, and it is stationary in front of decisions.
“Work in most processes is not slow. It is stationary.”
Why Does It Never Show Up in Your Metrics?
Here is the uncomfortable observation, and I intend it exactly as it reads: your operational dashboards are complicit. Enterprise metrics were built to measure activity, because activity is what costs visible money. Utilization, productivity, throughput per head, handle time: every one of them measures people being busy. A case aging six days in front of an approval generates no activity, so it generates no signal. The largest component of your cycle time lives in the gaps between everything you measure, and the dashboard reports a healthy, busy operation while customers experience an unresponsive one. We measure everything except waiting, and then we are surprised that waiting is where the value hides.
What Does Decision Latency Cost?
The cost is not abstract. Every day a case waits is a day of working capital financing it, a day of aging that breeds errors and rework, a day a customer spends forming an opinion of your responsiveness, and in penalty-bearing processes, a day of accumulating exposure. Latency also degrades the decisions themselves: information goes stale while the case waits, so the eventual judgment is made against conditions that have already moved. Slow decisions are not just late. They are frequently worse.
What Does Decision Latency Look Like in a Real Process?
Commercial construction offers as clean a specimen as any industry I know: the change order. A scope change on a major project touches the trade contractor, the general contractor, the architect, and the owner, each with review and approval rights. The engineering content of a typical change order represents hours of actual work. Its approval cycle routinely runs weeks, and on disputed items, months.
While it waits, the schedule absorbs risk, the contractor finances work it may perform at its own peril, and claims exposure compounds daily. Nothing about the documents is slow. Every day of that cycle is decision latency: parties waiting on assembled context, precedent terms, cost validation, and someone with authority to commit. The industry prices this latency into every bid, which means owners pay for it whether they see it or not.
How Do You Find Yours?
The instrument is a decision-latency audit: trace a sample of real cases end to end, timestamp every decision point, and attribute the waits. It is short, empirical work, and it reorders improvement priorities almost every time we run one, because the trapped value never sits where the org chart assumed. Structuring that analysis is part of our agentic AI consulting services, and it is the single fastest way to see your operation the way this series describes it.
You cannot manage a cost you refuse to measure. Measure the waiting, and the case for designing processes around their decisions stops being an argument and becomes arithmetic.
