How to quote lead times you can actually hit

Where the four-week answer comes from

Ask an estimator how the lead time on a quote was chosen and the answer is usually some version of the same thing. It is what was quoted last time. It is what the customer expects. It is what the owner said to say. None of those is a statement about whether the shop has the hours available.

This works until it does not. In a slow quarter the quoted date is conservative and work is turned away that could have been taken. In a busy quarter the same number is quoted while the floor is already three weeks oversold, and the date is missed. The shop then spends the following month expediting, running overtime, and explaining itself to a customer who is now shopping the next job elsewhere.

The underlying problem is that lead time is being quoted as a policy when it is actually a calculation. The inputs to that calculation already exist in the ERP. They are simply not being read at the moment the quote goes out.

What a real lead time is made of

A date the floor can hold has four components, and all four have to be present:

• Committed hours already on the books. Every released job has estimated hours on each work center. Those hours are a claim on future capacity whether or not anyone has added them up.
• Available hours by work center. Not shop-wide hours. A shop with plenty of welding capacity and a single overloaded CNC lathe does not have spare capacity, it has a bottleneck, and the quote has to reflect the bottleneck.
• The hours this job will add. From the routing on the quote itself.
• Material lead time. Which runs in parallel with the queue rather than adding to it, and is frequently the actual constraint on a job that everyone assumes is constrained by machine time.

Put those together and lead time stops being a policy. It becomes the answer to a specific question: given what is already committed, when does this job's last operation finish?

Building the load curve

The practical tool is a load curve: committed hours per work center, per week, forward for the next twelve to sixteen weeks, drawn against the hours that work center actually has.

Every job-based ERP holds the data for this. The job file has the routing. The routing has estimated hours per operation and the work center each operation runs on. The job has a due date. That is sufficient to spread committed hours across future weeks and see where the peaks are.

Three details separate a load curve that works from one that misleads:

• Use remaining hours, not estimated hours. A job that is eighty percent complete should contribute twenty percent of its hours to the forward load. Using full estimated hours on jobs already in process will overstate the load badly and push every quoted date out further than it needs to be.
• Use real available hours. A work center with two machines and one operator qualified to run them has one operator's worth of capacity, not two machines' worth. Hours available must reflect staffing and shift pattern, not the equipment list.
• Apply a utilization factor. No work center runs at one hundred percent. Setup, maintenance, breaks and the small interruptions of a working day mean that eighty to eighty-five percent of nominal hours is a realistic planning figure. Planning at one hundred percent guarantees the schedule is wrong.

Turning the curve into a date

With the load curve in place, quoting a date becomes mechanical. Walk the routing in sequence. For each operation, find the first week where that work center has enough uncommitted hours to absorb it. Move to the next operation and repeat from that week forward. The week the final operation completes is the earliest the job can ship.

Then add the two things the calculation does not know. Material lead time, if the material is not on the shelf, and a buffer. The buffer is a business decision rather than a mathematical one, but it should be explicit and consistent rather than invented per quote.

The output is a date with a reason behind it. When a customer pushes back on six weeks, the answer is no longer a shrug. It is that the machining operation cannot start until week three because the lathe is committed until then, and here is what moves if that job is expedited.

What changes once this is running

The obvious gain is that quoted dates get hit more often. The larger gain is less obvious and shows up in what the shop chooses to quote.

A load curve makes the shape of the next quarter visible. A trough eight weeks out is an argument for quoting aggressively on price to fill it. A peak in week four is an argument for quoting a longer date, or a higher price, or declining the work. Those are decisions a shop can only make if it can see the load, and most shops cannot.

It also changes the conversation about overtime and subcontracting. Both become planned responses to a visible peak rather than emergency reactions to a date that is already missed.

Starting without a project

This does not require new software. A shop can build a first version of a load curve from an ERP export into a spreadsheet in a day, and that version will be wrong in interesting ways that teach what the real constraints are.

The reason it usually does not stay in a spreadsheet is maintenance. A load curve is only useful if it reflects this morning's reality, which means it has to refresh from the ERP automatically rather than being rebuilt by hand each Monday. That is the point at which it becomes a system rather than a report, and the point at which people start trusting it enough to quote from it.