Skip to content

Systems and records

Commercial print proof approval data: approved on the wrong proof

By SourceX Editorial · Reviewed by Noah Loul ·

9 sections

Short answer

Represent each proof as a fixed version with an ID and file checksum, and tie the approval to that exact version rather than to the job. Then link the file released to plates or press, the press sign-off and the outcome records. A wrong-proof case is a mismatch between approved and produced versions, and its label is the recorded outcome, such as a reprint, a credit or acceptance.

Provenance trail

Commercial print proof

This page covers commercial print proof approval data: approved on the wrong proof.Illustrative

Key takeaways

  • An approval status on the job cannot show which proof version was approved; link approvals to version IDs and checksums.
  • Contract color proofs, content proofs and press sheet sign-offs approve different things, so record which one was signed.
  • Checksums show files differ; a content comparison shows whether the difference matters.
  • Separate approved errors from version mismatches, and label each case with the recorded outcome, not the job status.
  • Customer artwork and variable data often need withholding or redaction before proof history can be licensed.

How do you represent a print job approved against the wrong proof version?

Treat every proof as a fixed version with its own ID and file checksum, and link the approval to that exact version, not to the job. Then link the file the shop actually sent to plate or press, the press sign-off and the outcome records that followed, such as a reprint or a credit. A wrong-proof case is the mismatch between approved and produced versions, and its label is the decision the business finally recorded.

Most print management systems store approval as a status on the job: approved, yes or no. That field cannot show which version was approved, so a job approved on version two and printed from version three looks identical to a clean job. For commercial print proof approval data to teach or test anything, the version link has to be explicit and checkable.

This page describes the real proofing workflow, the fields a task packet needs, how to show that two versions differ, and how to decide the case. Example jobs and counts are illustrative.

Where the approved proof and the printed file drift apart

Drift happens in the gaps between systems. Customer files arrive by email, upload portal or file transfer; prepress preflights them, builds a soft proof or PDF proof and sends it out; the customer replies with changes; prepress makes a new version; eventually someone approves something. The approval, the file in the hot folder and the job ticket in the print MIS are three separate records that nothing forces to agree.

  • An approval arrives as a reply to an older email thread that carried an earlier PDF.
  • The customer approves in an online proofing portal while a corrected file is still being uploaded.
  • An approver signs a hard-copy contract proof after a later text correction was made only in the PDF.
  • Prepress pulls the right job number but the wrong file from a shared folder with several near-identical names.
  • A late change is approved by phone and recorded only in a CSR's note, not against a version.
  • Imposition or trapping creates a new production file whose content matches the approved proof but whose checksum does not.

Which proof is the authority for which decision?

Different proofs approve different things, so a packet must record which kind was signed. A color contract proof approves color appearance under a stated printing condition; a PDF content proof approves text, images and layout; a press sheet signed at the press approves the run against the proof. A job can be correctly approved for color and wrongly approved for content.

ISO standards separate these levels. ISO 12647-7 covers systems that produce hard-copy digital proofs intended to simulate a printing condition defined by characterization data, and ISO/TC 130 guidance describes these as contract proofs used in the contractual relationship between customer and printer and as a visual aim for the press. For the design stage, ISO 12647-8 covers a lower validation print level, without contract-proof quality.

Which proof is the authority for which decision?
Proof typeWhat a signature approvesEvidence to keep
PDF or soft content proofText, images, layout, page count and imposition intentVersion ID, file checksum, approval record with timestamp and approver
Contract color proofColor appearance under the stated printing conditionProof ID, simulated condition or profile, control strip result, signed copy
Validation printGeneral appearance during design, not final colorVersion ID and a note that it is not a contract proof
Press sheet sign-offThat the running sheet matches the approved proofSheet sample reference, time, signer, proof ID compared against
Change approval after proofA specific correction to a named versionChange description, new version ID, approver, how approval was given

The version authority and production decision task packet

Each packet covers one job and one approval decision, with every field filled from a named system. The template below is editorial, version 1.0 dated 2026-10-18; example values are illustrative and the system names are generic.

At a glance

The version authority and production decision task packet

The version authority and production decision task packet
FieldSource recordIllustrative valueCheck before use
Job and linePrint MIS job ticketJob 24-1187, brochure, 12 pagesTicket revision at time of approval is kept
Proof versionsProofing portal or prepress file logv1, v2, v3 with creation timesEvery version sent to the customer is listed
File checksum per versionPrepress file system or portalv2 and v3 hashes recordedHash computed on the exact file sent
Approval recordPortal approval, signed proof or emailApproved, v2, by customer marketing leadApproval points to a version ID, not just the job
Approver authorityCustomer contact list or order termsNamed approver on the orderApprover was entitled to approve for the customer
Released production fileRIP or platemaking logFile with v2 hash sent to platesReleased hash matches an approved version
Press sign-offPress OK sheet recordSigned sheet compared against v2 proofSign-off names the proof it was checked against
DiscrepancyComplaint, CSR note or quality reportWrong phone number found at deliveryDiscovery date and who found it
Decision and outcomeCredit memo, reprint job, customer acceptanceReprint job 24-1203 openedOutcome record postdates the discrepancy
Counterexample linkPaired job with checksum change onlyRe-imposed file, identical contentContent comparison result is stored

What an approval record must contain to be usable

It must name a state, a person with a role, and the file. The CIP4 XJDF schema, a job ticket format for print production, models this directly: an ApprovalDetails resource carries a required ApprovalState of Approved, ApprovedWithComment or Rejected, and can carry the approving person, a comment and a FileSpec pointing to the file, while each ApprovalPerson references a contact and can have the role Approvinator, Informative or Obligated.

The FileSpec element in the same schema includes optional CheckSum and URL attributes, which is the link most shops lose. Even if a shop never exchanges XJDF, these fields are a good checklist for its own export: state, approver, role, comment, file location and checksum.

Treat approved-with-comment as its own case. The comment may ask for a change that should have produced a new version needing approval, or it may be a remark that left the file untouched. The packet should record which, because that decides whether production on the commented version was correct.

Proving that two proof versions really differ

Compare checksums first, then content. Different checksums prove the files differ in some byte, but re-saving, imposing or adding printer marks changes the checksum without changing what the customer approved. Same checksums prove the files are identical.

Where checksums differ, a content comparison settles it. Adobe Acrobat's Compare Files tool, for example, compares two versions of a PDF and produces a report that opens with a summary of changes, and it can be limited to certain pages or to text only. Store the comparison output in the packet so a reviewer can see the actual difference, such as a changed phone number, without opening both files.

File format affects how far a checksum can be trusted. Under ISO 15930-7, a PDF/X-4 file contains everything needed for final print reproduction, while a PDF/X-4p file relies on an externally supplied, unambiguously identified ICC profile. For partial-exchange files the packet must record the external profile as well, since a different profile can change the printed result while the PDF checksum stays the same.

Which condition decides whether the obligation was met?

The decision turns on two facts: which version the customer approved, and which version the shop produced. If production matched the approved version and the error was in that version, the case is an approved error. If production used a different version from the one approved, the case is a version mismatch, regardless of which version contained the error.

How responsibility is split in each case depends on the shop's quote, order and standard terms in force for that job, so the packet should include the relevant clause text or reference rather than encode a general rule. Record also whether the customer accepted the printed job as is, asked for a reprint, or was given a credit, because that recorded outcome, not the job status, is the label.

Keep cases where the evidence is incomplete, such as an approval by phone with no version reference, but flag them as unresolved. A model that sees only clean cases learns to assume every approval points to a version.

The same packets support a narrower and more useful task than deciding blame after the fact: checking a job before release. Given the approval record, the version list and the file queued for plates, a system should say whether the queued file matches an approved version, whether any correction was posted after the approval, and whether the approver was the one named on the order. Score it on whether it would have stopped the mismatched jobs and let the counterexamples through, and keep the reasons it gives, since a hold that cites the wrong version is no better than no hold.

Sample packets across job types, customers and proofing channels, and include a share of uneventful jobs. If every packet in a set ended in a reprint, a system learns that disputes always end that way and never sees the ordinary approvals that make up most of a shop's work.

Illustrative case: a corrected phone number that never reached press

Illustrative scenario with invented details. A 12-page brochure goes through three PDF proofs. Version two fixes an image; version three corrects a phone number on the back page. The customer replies "approved, go ahead" to the email that carried version two, two hours after version three was posted to the proofing portal.

The print MIS job ticket shows approved. Prepress, working from the email, pulls version two from the job folder and releases it to plates; the RIP log records its checksum. The press operator signs the OK sheet against the version two proof printout. The job ships, and the customer reports the old phone number on delivery.

The packet links the approval to version two by the email attachment checksum, the released file to version two by the RIP log, and the comparison report shows the single text change between versions two and three. Decision in this illustration: production matched the approved version, so this is an approved error, but the shop's own procedure required re-approval after any posted correction, so the job is flagged for a process exception. The outcome record is a reprint job ticket and a customer acceptance note.

The counterexample is a similar job where version three only re-imposed the pages for a different press. Checksums differ, the content comparison finds no text or image changes, and printing from version three is correct.

From proof history to a licensable record set

Proof histories are useful records for AI work because they connect intent, approval and outcome, but they carry other people's content. Customer artwork, logos, copy and photographs usually belong to the customer, and variable-data jobs can contain names and addresses. A shop can often keep job metadata, version links, comparison results and outcomes while withholding or redacting the artwork itself, and that choice should be made job type by job type with the customer terms in view.

Access is not the same as permission. The shop holding the files, the privacy position of any personal data on them and the right to license them are separate questions; de-identification lowers privacy risk but does not create a licensing right where customer terms restrict reuse.

SourceX licenses selected operational records directly from print businesses and runs sourcing, rights review, contracting, preparation, delivery and payment. The shop approves scope and terms and signs before any job record is delivered, de-identification is then handled by SourceX through approved processing routes, and a short assessment with no data uploads shows whether a printer's records may qualify.

Frequently asked questions

What is commercial print proof approval data?

It is the record of every proof version sent to a customer, the approvals and rejections against each version, the file released to production and what happened after delivery. Useful data links these by version ID and checksum rather than storing approval as a single status on the job.

How do you prove which proof version a customer approved?

Link the approval to a version ID and the checksum of the file the customer saw, for example the attachment on the approving email or the file shown in the proofing portal. If the approval names only the job, record it as unresolved rather than guessing.

Is a different file checksum enough to show the wrong proof was printed?

No. Imposition, printer marks or re-saving change the checksum without changing approved content. Run a content comparison between the approved and produced files and keep the result with the case.

What is the difference between a contract proof and a content proof?

A contract proof, covered by ISO 12647-7, is a color proof that simulates a defined printing condition and serves as the color reference between customer and printer. A content proof, usually a PDF, approves text, images and layout and says little about final color.

Can a printer license proof history if the artwork belongs to customers?

Often only in part. Job metadata, version links, comparison results and outcomes may be usable, while artwork and personal data in variable-data jobs may need to be withheld or redacted. Customer terms and a rights review decide what can be included.

Sources

  • ISO 12647-7:2016 specifies requirements for systems that produce hard-copy digital proof prints intended to simulate a printing condition defined by characterization data. Source
  • ISO/TC 130 guidance describes ISO 12647-7 digital proofs as contract proofs used in the contractual relationship between customer and printer, and ISO 12647-8 validation prints as a design-stage level without contract-proof quality. Source
  • ISO 15930-7 designates a file containing all elements needed for final print reproduction as PDF/X-4, and one relying on an externally supplied, unambiguously identified ICC profile as PDF/X-4p. Source
  • The CIP4 XJDF schema defines ApprovalDetails with an ApprovalState of Approved, ApprovedWithComment or Rejected, ApprovalPerson roles, and a FileSpec with CheckSum and URL attributes. Source
  • Adobe Acrobat's Compare Files tool compares two versions of a PDF and opens a results report with a summary of changes, optionally limited to pages or text only. Source

See if your company qualifies

A short company assessment. No data uploads are needed.

See if your company qualifies