In the canonical flow walk, the enterprise begins at an indexed 100 units of potential throughput, drops to 84.6 once planning and materials have taken their cut, and ends at 64.2, with more than a third of the potential lost between functions that each report healthy local numbers. This is the signature of local optimization: a set of green dashboards summing to a red enterprise.
Everyone is right and the enterprise is late
Each function is measured on the thing it controls: utilization, cycle time, cost per unit. Each behaves correctly against that metric. Manufacturing protects its utilization. Logistics protects its cost. Sales protects its close rate. None is wrong, and the enterprise still misses, because value is created vertically inside functions and lost horizontally between them. The handoff, the interval where one function’s output becomes another’s input, is owned by no one, measured by no one, and therefore defended by no one.
The awareness gap is not the problem; the accountability gap is. The American Management Association finds 83% of executives recognize silos in their organization and 97% say they harm business outcomes, yet the losses persist, because recognizing a silo and owning the space between two of them are entirely different things.
Every function is right. The enterprise is still late. The interval between them belongs to no one.
The constraint starves while departments protect utilization
When every department optimizes its own utilization, the enterprise constraint is starved by design. A non-bottleneck running at full utilization does not create throughput; it creates inventory and queues in front of the actual constraint. This is why adding capacity to a green department rarely helps, the bottleneck is somewhere else, and the local metric that looks best is often the one feeding the problem. McKinsey estimates friction consumes 20–30% of organizational capacity in siloed companies; BCG puts redundant-activity loss at up to 15% of workforce capacity. That is capacity paid for and not converted to output.
Why local metrics conceal it
The elapsed interval the customer actually experiences, total lead time from order to delivery, is the sum of the work plus every handoff between the steps. Local metrics measure the work and ignore the interval. One analysis of a single product-launch process found 41 handoffs, averaging 3.7 days of delay each, at roughly $27,000 per handoff, none of which appeared on any department’s scorecard. The composite lead time exceeded the work it contained by a wide margin, and no function’s report showed it, because no function was measured on it.
The recovery mechanism
The fix is threefold and sequential. Schedule against the actual enterprise constraint rather than departmental utilization. Measure the interval between functions, the handoff itself, rather than the steps inside it. And change what departments are optimized for, because sequencing follows measurement: as long as local utilization is the target, local utilization is what you will get. End-to-end flow is the only thing that improves end-to-end, and it improves only when someone finally owns the white space between the boxes.