A promise date from standard times is a guess with a decimal point

A customer asks when. Somebody opens the schedule, finds a gap, adds the standard times for each operation, adds a safety margin that was set years ago, and gives a date.

That date has three decimal places of apparent precision and is built on a number describing an operation that runs cleanly, with material present, an experienced operator, no changeover, and nothing going wrong. Real production is the same operation plus everything that interrupts it.

What standard times leave out

Changeovers. Often the largest single consumer of capacity in a plant making varied products, and frequently held as an average rather than as the sequence-dependent number it actually is.

Micro-stoppages. The two-minute clear, the jam, the adjustment. Individually invisible, collectively a large fraction of the shift, and almost never logged because nobody stops to write down two minutes.

Waiting. For material, for an operator, for the previous operation, for a quality release. This is usually the majority of the time an order spends in the plant, and it appears in no routing.

Who is on. The same operation runs at different speeds depending on who is running it, which everybody on the floor knows and no system records.

Quality. First-pass yield below a hundred per cent means some units go round twice. The schedule assumes they do not.

Why the margin does not fix it

Plants know their standards are optimistic, so they add a buffer. The buffer is a single number applied everywhere, which means it is wrong in both directions at once: too generous on straightforward orders and nowhere near enough on the difficult ones.

The result is a schedule that is simultaneously too slow to be competitive and too optimistic to be reliable, which is the worst of both and is where most plants live.

What a real date is built from

Achieved throughput, per product, per line. What this line actually produces when making this thing, measured rather than assumed.

Changeover from the sequence. What it costs to go from what is running now to what is next, which is a matrix rather than a constant.

The queue in front of the constraint. Most of an order’s time is spent waiting for one resource. A date that ignores the queue is a date for an empty plant.

A range, not a point. Real production has variance. A range with a confidence attached is more honest and more useful than a single date that everybody privately discounts.

The uncomfortable part

Dates built from real data are frequently later than the ones being quoted today, and telling sales that is not a popular conversation.

But the current dates are not earlier in any sense that matters — they are the same dates, missed. Quoting three weeks and delivering in five is worse for a customer than quoting five, because they planned around three. The plant that can quote honestly wins the orders it can actually deliver and stops winning the ones that were always going to end badly.

That is not a reporting improvement. It is the difference between a schedule people believe and one everybody privately adds two weeks to.