Preparation gets described as tidying up. It is not. It is the stage where a physical arrangement of paper, full of implicit information about what belongs with what, is translated into something a machine can process without losing that meaning. Get it wrong and the images will be perfect and the archive will be unusable.
Decide these before a single box is touched
- What counts as one document. A file? An invoice with its attachments? Each sheet? This rule determines where files split and where index data attaches, and changing it later means reprocessing.
- Which index fields you need, and where each value is found on the page. If a field is not printed on the document, say now where it comes from: a file cover, a register, a folder label.
- What happens to the paper afterwards: return, storage or destruction. This changes whether files must be re-fastened in original order, which changes how preparation is done.
- Whether original file order carries meaning. In some archives, sequence is the only chronology you have.
- Who answers questions. Preparation generates queries every day, and unanswered ones become an exceptions pile.
- What must not be scanned. Personal items, duplicates, superseded drafts, records already past retention. Decide before, not during.
The physical work, item by item
Fasteners
Everything metal or plastic comes out: staples, paper clips, bulldog clips, treasury tags, binder rings, plastic combs, spiral bindings. A single missed staple can jam a feeder, tear pages and stop a line. Use a proper staple extractor rather than pulling by hand, because torn corners create double-feed errors later. Where a fastener held documents in a meaningful group, mark the group with a separator sheet before you remove it, or that grouping is gone.
Sticky notes and small inserts
Leave sticky notes in place but flag them, because they usually carry information the scan would otherwise lose: an approval initial, a query, a corrected figure. The two workable approaches are to move the note onto a blank carrier sheet placed immediately behind the page it came from, or to scan the page twice, once with the note in position and once without so the covered text is captured. What you must not do is peel notes off and bin them, which is exactly what an unbriefed helper will do. Small receipts and slips get the same treatment: mounted on a carrier sheet, in place, not loose in a stack.
Damage
Repair torn edges with archival tape before capture, not after a page has been shredded by a feeder. Flatten folded corners and creases; a fold across text produces a shadow line straight through the characters and degrades OCR. Brittle, damp-damaged or very thin pages should be pulled out and marked for flatbed capture rather than fed through a roller. Set them aside as you find them instead of hoping the operator notices.
Double-sided pages
Scan duplex by default and let blank-page detection remove the empty backs. The reason is simple: a stack that looks single-sided almost always contains some sheets with a signature, an amendment or a stamp on the reverse, and a single-sided pass loses them silently. Blank detection needs a tolerance setting, because a page carrying only a faint stamp can be discarded as blank. Where the back of a page matters, flag it rather than relying on the setting.
Odd sizes and orientations
Sort oversized sheets, folded plans, receipts and anything below feeder minimum into their own batches. Mixing sizes forces reconfiguration mid-stack or produces cropped edges. Note pages printed landscape so they can be rotated consistently at capture rather than corrected one at a time afterwards.
Separator sheets: where documents get their boundaries
A separator sheet is a page inserted into the stack that tells the capture software a new document begins here. It is usually a barcoded sheet, sometimes a coloured one, and it is how a continuous run of images becomes a set of discrete documents.
- Insert a separator at every boundary defined by your document rule, not just at obvious file breaks.
- Use different separator types for different levels: box, file, document. This is what lets a hierarchy be rebuilt automatically.
- Where a separator also carries index data, such as a barcode encoding a file reference, it removes keying work and reduces error. This is the single highest-value preparation refinement available on most projects.
- Count your separators against your document count. A mismatch is the first sign that a batch will split wrongly.
- Never reuse separator sheets between batches without checking they were removed. A stray separator mid-file splits a document in two, and the halves get different index values.
Labelling boxes so the index means something
Box labelling looks administrative and is actually the backbone of reconciliation. It answers the question everyone asks at the end: is everything here?
- One unique identifier per box, applied physically and recorded in a manifest at the same moment.
- Record the department, record category and date range for each box. These become the fallback index for anything that cannot be identified from its own content.
- Record a count: files, or documents, or an estimated sheet count taken the same way for every box. Consistency matters more than precision here, because this is what the final reconciliation compares against.
- Note anything unusual on the label: mixed categories, known gaps, damaged contents, items that must be returned separately.
- Keep the manifest outside the boxes. A manifest that ships inside the container it describes has no value if the container is the thing that goes missing.
What not to do
- Do not re-file or reorganise while preparing. If original order carries meaning, reorganising destroys it, and if it does not, reorganising is wasted effort.
- Do not photocopy faded originals to make them clearer. A copy of a faint page is worse than the original, and greyscale capture at a higher resolution nearly always beats it.
- Do not discard anything without an instruction, including duplicates. Two copies with different handwritten annotations are two documents.
- Do not write on originals to mark boundaries. Use separator sheets.
- Do not remove documents from the batch to answer a query and then return them somewhere else in the stack.
- Do not let an unbriefed team start. One afternoon of enthusiastic help that strips every sticky note and re-sorts alphabetically cannot be undone.
Frequently asked questions
Should we remove staples ourselves before collection?
You can, and it may reduce the quoted preparation effort, but only if done to the supplier's standard. Torn corners from hand-pulled staples cause double feeds and rework that costs more than the saving. Ask for the written preparation specification, trial one box, and have it checked before committing your team to the whole archive.
What happens to sticky notes during scanning?
They should be preserved, not discarded, because they often carry approvals, corrections or queries found nowhere else. The usual methods are mounting the note on a carrier sheet placed directly behind its original page, or scanning the page twice, with and without the note, so the covered text is also captured. Agree the method before preparation starts.
Do documents need to be in a particular order?
They need to be in a known order, and the safest one is the order they are already in. If existing sequence carries meaning, such as chronology in a case file, preserve it exactly and say so in the brief. Reorganising during preparation is the most common way archives lose information that no index can restore.
How are torn or fragile pages handled?
Repair tears with archival tape before capture, then route fragile, brittle or damp-damaged pages to flatbed scanning rather than an automatic feeder. Identify and separate them during preparation instead of relying on an operator to spot them mid-stack. Very fragile items may need to be handled individually, which should be flagged at assessment so it is scoped and priced.
What is a separator sheet and do we need one?
A separator sheet is a page inserted into the stack, usually barcoded, that tells the capture software where one document ends and the next begins. Without them, a batch arrives as one long image sequence that has to be split by hand. If a separator can also carry a file reference as a barcode, it removes keying work and cuts indexing errors.
About this article
Written and reviewed by the digitization delivery team at Document Digitization Services, the specialist division of Athena Global Technologies LLC. Content is reviewed against how projects are actually run, and updated when that changes.