Repeat orders: what makes the second run cheaper?

What changes on a repeat order, and what does not?

A customer reorders a part you have made before. The easy move is to take last time's price from the ERP. That figure does not say why the price was what it was. It may hold a fixture, a full first article inspection or a material price from last year.

Some of the work already exists: the NC program, the fixture, the setup sheet and, if you booked them, the actual times from the last order. Other things move: the purchase price of the material, sometimes the drawing revision, sometimes what the customer wants in inspection. The table sets them side by side.

Which one-off costs drop out?

One-off costs come once per order, whatever the quantity: programming, making a fixture, setting up and measuring the first part. On a small batch they make up much of the unit price; why a small batch costs so much per part shows how much.

On a repeat order at the same revision, most of the programming time goes. The program has run and the first parts were approved. What is left is loading it and checking the tool lengths, and that belongs to setup.

The fixture drops out if it was paid for on the first order and is still on the shelf. Agree at the first quote whether a dedicated fixture is invoiced separately or built into the unit price, and who owns it afterwards. Leave that open and the repeat order charges for it twice, or not at all.

What does setup cost the second time?

Setup does not vanish: the machine still has to be changed over from the last job. The Lean Enterprise Institute's lexicon counts changeover time from the last piece of the previous run to the first good piece after the changeover.

The second time is shorter because the steps are known. A setup sheet records which fixture, which work origin and which tools belong to the job. SMED splits setup into work that needs the machine stopped and work that can be done while it runs, and moves as much as possible into the second.

Zero-point clamping takes this furthest. The fixture sits on a pallet or base plate with fixed reference points, so the work origin lands in the same place after every change. In 2014 the Dutch trade paper MetaalNieuws described a shop that used a zero-point system to clamp parts onto pallets outside the machine and cut its daily setup from hours to minutes.

So do not copy the first order's setup time. Take it as the upper limit and estimate how much of it was preparation that is now ready: building the fixture, finding the work origin, proving out the program. How the number of setups drives that time is in count the setups before you name a price.

Does the revision still match?

A repeat order is only a repeat if the drawing is the same. Compare the revision letter or number with the last order's before you reuse anything. A new revision means a drawing change, and a change that looks small can still alter the route.

A tighter tolerance on one bore can call for an extra finishing pass or a different machine. A different finish, anodised instead of bare, brings a subcontract line and sometimes a dimensional allowance. A changed thread needs a different tool and sometimes a new program. The last estimate then holds only for the lines the change does not touch; a revision without a change note is checked dimension by dimension.

Worked example: a first order and a repeat of 25 parts

Round numbers: a milled part, 25 per order, same revision. Material cost € 8 a part at the price date of the first quote and costs € 9 now. Cycle time costs € 12 a part and is held the same on both orders.

The unit price almost halves, from € 52.80 to € 26.60. The one-off lines fall by € 680. Material goes the other way, by € 25 for the whole order. Copy € 52.80 and the customer pays for fixture and program twice; halve it by feel and you miss the dearer material.

Cycle time stays the same in this example. If you booked the actual time per operation on the last order, that measured time replaces the estimate.

What do you learn from last time?

The last order is the best evidence you have, provided you keep the actuals per operation and not just per order. A Subduxion guide shows that only a comparison per operation reveals where an estimate went wrong.

Check, too, whether the booked time was representative. A broken cutter or a breakdown makes a time longer without the operation being any different. An outlier like that leaves the standard alone. If an operation ran over without such a cause, correct only that operation's standard and write the reason down. If the same gap comes back over several orders, it is a signal, not chance.

An ERP keeps last time's price and usually the booked hours, but not why the price was what it was: which line was one-off, which time ran over, and why. Those reasons belong with the part, not in one cost engineer's head. How to set estimate and actual side by side is in estimate versus actuals per operation.

How to estimate a repeat request

Go through these questions before you name a price, in this order.

• Is the revision the same as on the last order?
• Do the program and fixture still exist, and where are they?
• Is there a setup sheet, and was it right last time?
• What was the actual time per operation, and why did it differ?
• What is the material purchase price today?
• Does the customer ask for a full first article inspection again?
• Is the quantity different from last time, and does that change the quantity break?