Approvals, and who pays for the reprint

The question every dispute reduces to

Five thousand copies are wrong. Somebody pays. The only question that matters is whether the supplied file was wrong or the press deviated from it, and it is answerable only if you kept the evidence.

What to keep, on every job

• The exact PDF you sent. Not “the final version”: the actual bytes. Archive it alongside the job, and record a checksum:

shasum -a 256 job-press-v3.pdf > job-press-v3.sha256

Thirty seconds, and it converts “this is the file I sent” from an assertion into a fact.

• The email or message that sent it, with the spec paragraph in the body.
• The proof, signed and dated, or the email in which the client approved.
• Your preflight output, saved as a text file.

Four artefacts. Together they answer the question before it becomes an argument.

Approving a proof properly

1. View under a known light. D50 if you have it, north daylight otherwise. Never a warm tube.
2. Compare against the reference, not against memory or a screen.
3. Read every word out loud. Names, phone numbers, prices, dates, the word “public”. Have a second person read it independently.
4. Mark corrections physically, in a colour not present in the job, with a number beside each so they can be counted.
5. Sign and date it, photograph it, and send the photograph back so both parties hold the same record.

What a signature means

Signing a proof normally means: the content is correct and I accept this as the standard for the run. It transfers responsibility for anything visible on the proof.

So read what you are signing for. A digital proof shows content, position and approximate colour. It does not show the stock, the lamination, the fold, the trim accuracy or the exact ink density. Signing one does not mean you accepted a 4 mm trim drift.

Where colour matters and the budget allows, ask for a press pass: be there when the run starts, look at the first good sheet, and sign that. It is the strongest approval available and it costs an hour.

The disputes that actually happen

Colour is not what I expected. Answer: what was it compared against? If a printed reference was agreed and a ΔE tolerance was named, this is measurable. If the reference was a screen, nobody can win, which is why the colour module insisted on a printed reference.

Trim is off. A 1 to 2 mm drift is within normal guillotine tolerance and is why safe areas exist. A border sitting on the trim line will look uneven and that is a design decision, not a press fault. Beyond about 2 mm, it is worth raising.

A typo. If it was on the signed proof, it is the approver’s. This is uncomfortable and it is why the read-aloud step exists.

The file was rejected or changed. If a shop altered artwork (converted a spot, substituted a font, rebuilt a die-line) ask for the file they output. A reputable shop will send it.

It came back late. Agree the schedule in writing including who does what by when, because a proof sat on for four days is not the printer’s delay.

How to handle it when it is your fault

Say so quickly, in plain words, and propose a remedy before you are asked. “The die-line layer was set to knock out rather than overprint. That is my error. I will supply a corrected file today and I will cover the plate cost.”

This is not only decent; it is commercially sound. Printers and clients both remember who was straight with them about a mistake, and in a trade that runs largely on repeat work and word of mouth, that memory is worth more than the cost of a set of plates.

Setting expectations at the start

Two sentences in your quotation prevent most of this:

Proofs will be supplied for approval before printing. Changes after approval are chargeable, and colour is matched to the supplied printed reference within normal commercial tolerance.

Say it once, in writing, and the awkward conversation mostly does not happen.