# What exposes you on inspection is not the removal. It is the missing file. Two hundred documents can become twenty without a single requirement ceasing to be covered. **On one condition, and it is a complete one**: being able to show, for each of the one hundred and eighty withdrawn, what it contained, where that content now lives, who decided and on what grounds. Withdrawn ≠ erased Document control requires superseded versions to be removed from points of use _and_ retained in archive. A document taken off the workstation remains consultable. These are two distinct obligations, and they do not conflict. Two hundred documents in force become twenty; the other one hundred and eighty go to archive.200in force20in force180archivedWithdrawn from use, retained and retrievable. A principle common to the GMP documentation chapter and to document control under ISO 13485. That distinction dissolves most of the anxiety. **Nobody is asking you to keep in force what no longer serves; you are asked to know what existed, and until when.** A lighter, properly archived document set satisfies the requirement better than a sprawling one in which nobody can guarantee that no obsolete version is still circulating at the workstation. "If someone wrote it one day, there must have been a reason." Probably — and that is exactly the question to settle. **In a document set that has aged, a notable share of documents answers a requirement that has gone, equipment that has been dismantled, or an audit observation closed ten years ago.** The reason existed; it no longer does. Not daring to check means keeping in force texts nobody can any longer attach to anything — and in front of an inspector, that inability is a weakness, not a protection. ## Four items in the file. ### 1\. The dated baseline A photograph of the document set before you start: full inventory, versions in force, last review dates, consultation rates. It is the reference everything afterwards is compared with. Without it no demonstration is possible — and it cannot be reconstructed later. ### 2\. The coverage matrix For each applicable requirement, the document or documents covering it, before and after. This is the centrepiece: it proves no requirement has been orphaned. It also reveals, nearly always, two or three requirements the original set did not cover — the exercise improves compliance, it does not merely lighten it. ### 3\. The decision, document by document Merged into which document, withdrawn on what grounds, replaced by what. One line is enough, but it must exist for every one. That register is what turns a clean-up into an act of quality management — and it reads in ten minutes in the room. ### 4\. Going through change control Impact assessment, approval at the right level, update of the associated qualifications, archiving of withdrawn versions. Nothing exotic: the loop you already have, applied to a documentation operation. That is where the line runs between simplifying and deviating. ## The inspector is not an adversary, provided you hide nothing. A well-kept simplification file is one of the best interview topics a site can offer: it shows an organisation that knows its document set, knows what it covers and why, and decides rather than accumulates. **Explaining is part of the file** — and it is that, far more than volume, which builds confidence. ## A document set you can actually hold. Review Periodic review workload is proportional to the number of documents in force. Halving the set halves an annual burden that falls on the same people every year. Coherence Fewer texts means fewer chances that they contradict each other. A contradiction between two documents in force is one of the hardest findings to explain away in the room. Deviations On the sites we support, the share of deviations of documentary origin falls markedly — the most tangible effect, and one you read in your own indicators. So the programme does not start with rewriting, but with measurement: **the dated baseline is both the first item in the file and the best argument for launching the rest.** ## Go further. [ ### The document lifecycle Over-revision and fossilisation. Read →](/en/blog/quality-document-management-lifecycle/)[ ### Simplify without deviating Where the line runs exactly. Read →](/en/blog/gmp-simplify-without-deviating/)[ ### Preparing for an inspection When the system is fragile. Read →](/en/blog/prepare-inspection-fragile-system/) ## Simplification, plainly. How long must a withdrawn document be kept? + The period depends on the nature of the document and on your archiving policy, itself derived from the requirements applicable to your products — shelf life, local regulation, contractual commitments. The principle does not vary: withdrawn from use, retained in archive, retrievable. Your policy sets the number of years, not the simplification programme. Do you need one change control per document removed? + Rarely, and it would be unmanageable. One change control per coherent batch — a process, an area, a document family — with the line-by-line decision register in annex, is both compliant and workable. What matters is that every withdrawal is justified and approved, not that it has its own form. How do you detect duplicates and contradictions? + By hand, it is several weeks of work on a set of a few hundred documents, and it is rarely exhaustive. Automated content analysis, from a simple document management export, produces the list of overlaps and divergences overnight. The decision remains entirely human. And if an auditor asks for a removed document? + You retrieve it from the archive, you show the register line giving the date and grounds for withdrawal, and you show in the coverage matrix where its content now lives. That sequence takes two minutes and closes the topic. It is precisely what the file exists for. ## Do you know which requirement each of your documents answers? Two days on an export of your document system is enough to produce the dated baseline — the first item in the file, and often the most convincing.