Where is enterprise flow breaking down · and which bottlenecks are actually limiting your performance?
Capacity is signed off asset by asset, and system output is never measured against the sum. On a $210M capacity footprint, the throughput walk traces 100% of available capacity down to the 64.2% that physically ships (three zone yields of 84.6%, 84.1% and 90.3% that multiply rather than average) names every constraint in between, and separates the $37.24M that is recoverable from the $37.89M that is the cost of operating a real plant.
Capacity is measured asset by asset. Nobody measures the distance between them.
Every asset can run at its target and the enterprise can still ship two thirds of what it is paying for.
That distance is what the Flow Constraint Analyzer measures.
Six functions each protect a number. None of them is what leaves the dock.
Every one of these measures is legitimate, and every one of them can be hit in full while system output falls. Throughput is not lost to weak departments. It is lost in the handoffs between strong ones, where no metric is defined.
Accuracy at the aggregate level, refreshed on a cadence the plant cannot schedule against.
Price against standard, with lead-time variance carried by whoever is running the line.
Absorption and run rates, which reward batch sizes that starve the true constraint.
PM compliance across the whole estate, weighted equally rather than by constraint impact.
Defects caught at inspection, after the constraint has already spent its hours making them.
Dock productivity and freight rate, paid for in finished goods aging in the yard.
Capacity is a per-asset number. Throughput is what survives every handoff between them.
Theoretical capacity assumes perfect materials, a stable schedule, no unplanned stops, no rework and no queue at the dock. Every handoff between those conditions carries its own yield, and yields do not average, they multiply.
Fill rate and schedule stability, before a single machine starts.
Equipment reliability and first-pass yield on the floor.
Fulfillment velocity from finished goods to the customer dock.
What physically leaves the dock against what the plan committed.
Three zones each running in the mid-eighties compound to a system running in the low sixties. That is arithmetic, not underperformance, and it is why one more machine bolted onto a mid-eighties zone changes almost nothing.
Five constraint classes, traced from available capacity to the customer dock.
The walk reads the same production hour from both ends, what the plan committed, and what physically shipped against it.
Fill rate against the schedule, and the runs that started short or did not start at all.
Churn between plan release and execution: resequencing, expedites, and the changeover minutes each one costs.
Unplanned stops weighted by constraint impact, not spread evenly across the asset base.
Rework that consumes constraint capacity twice, once to make the defect, once to correct it.
Finished goods queuing for waves, slots and tenders, throughput won on the floor and surrendered in the yard.
Raw execution records to a board-ready throughput walk, in one week.
Data request and validation. Work order, movement and downtime records, read-only.
Execution ingestion. Production, material and shipment records stream as they exist today.
Multi-agent reasoning across planning, execution and fulfillment records against sector baselines.
Throughput walk delivered, every constraint sized and attributed to the function that produced it.
Available capacity, decomposed to what physically left the dock.
The same walk the analyzer below produces at its default settings. Five constraint classes, each attributed to the function whose decision created it, then the portion that is genuinely recoverable, separated from the portion that is the cost of operating a real plant.
That was a reference footprint. The next figure on this page is yours.
That was a $210M footprint. Put your own numbers in.
The same five constraint classes, driven by your scale and your execution discipline. Move one slider at a time and watch what it does to the net yield on the right. That is the whole argument for reading capacity horizontally instead of asset by asset. Nothing is transmitted; it runs entirely in your browser.
01 · Enterprise scale
02 · Pre-production flow
03 · Execution & fulfillment
Two ways to find the same constraint.
- Time-and-motion sampling on a handful of lines
- Manual reconciliation across MES, ERP and spreadsheets
- Findings presented at the end of the engagement
- 30+ hours of planning and operations interviews
- Every work order, not a sampled line
- Automated multi-agent reconciliation
- The figure stated before the engagement begins
- Zero personnel friction · direct log interaction
Where capacity is lost, how much is recoverable, and which function owns each point.
Theoretical capacity decomposed to actual shipped volume, every constraint named and sized.
Chart · WalkThe three compounding zone yields, so leadership can see which one is actually setting the ceiling.
Chart · ChainBottlenecks ordered by financial impact rather than by how loudly they are reported.
Table · RankedEach lost point traced to the decision and the function that produced it.
Matrix · By functionWhat the recoverable capacity is worth against the capex request it makes unnecessary.
Figure · DeferralThe instrument that found the constraint is the one that watches it.
A recovery that no one measures afterward is a recovery you will run again in eighteen months. The instrument that produced the number is the instrument that watches it: same walk, same segments, same attribution: refreshed, so the only question left is whether each line is improving, regressing, or flat.
Watches for: Planning cadence slippage, setup-family erosion, constraint-asset downtime, dock queue growth
Illustrative twelve-month deployment, including a realistic regression at month five when the new planning cadence meets the old incentives. Not client data.
You cannot buy your way out of this.
A capital request adds capacity to one zone. The system multiplies three of them, so the money buys a fraction of what the business case promised, and the constraint moves rather than clears.
Pressure-test the 64% question against your own plan-to-ship reality.
Twenty minutes with an xChangeFlow principal. Bring skepticism; we'll bring your industry's baselines. You leave knowing whether a seven-day analyzer run is worth your data export.