Digitizing a paper archive: a step-by-step plan

Short answer: digitizing a paper archive is a project in which paper documents are prepared, scanned, turned into searchable text and described with metadata, so the electronic copy is found in a moment and you always know where the paper original is. A good plan runs in this order: scope and priorities, description rules, paper preparation, scanning, OCR, classification, quality control and a decision on the originals. In NG DMS one system covers the whole path from scan to archive.

What is archive digitization, and what do you gain?

Archive digitization turns paper documents into electronic ones that can be searched, described, shared under permissions and kept with proof of integrity. A scan on its own is not the goal: an image of a page with no text and no description is just a heavier binder. The value appears when the scan gets text, a type, a correspondent, a date and a link back to the paper original.

Done properly, it means:

• a document is found by a word on the page, a number or a correspondent, not by its shelf;
• several people work with the same document at once, without borrowing the binder;
• who opened or changed a document shows in the trail;
• scanned documents can enter approval workflows, the records registry and NG Copilot's answers.

Which digitization approach should you choose?

Before switching on the scanner, decide what gets digitized and in what order. Four approaches are common:

A common and sensible mix: new mail from today, the existing archive by category, and the rest on demand.

A digitization plan in eight steps

1. Scope and priorities — what to scan, in what order, and what to skip.
2. Description rules — types, required fields, correspondents, confidentiality.
3. Paper preparation — staples, page order, labels and separators.
4. Scanning — settings, the channel into the system and a pilot.
5. OCR and archival copy — text for every page and PDF/A-2b.
6. Classification and metadata — the system suggests, a person confirms.
7. Quality control — a sample from every box, not at the end.
8. Digitization record and originals — proof of the procedure, and where the paper is.

1. Scope, priorities and retention periods

List what you have: which kinds of documents, from which years, how many binders and boxes, and where they are. Compare the list with your list of record categories and retention periods. Material whose retention period has expired is usually not worth scanning — there is a disposal procedure for it, carried out only with the competent archive's decision (see The archive book and the list of record categories). For material kept permanently, check the rules on keeping originals with the competent archive in advance.

2. Description rules before the first scan

The most expensive mistake in digitization is to scan first and think about description later. Before the first stack, decide:

• which document types you have and which fields are required for each, such as contract number, date and amount;
• who your correspondents are and how they are kept — the register can be imported from CSV or Excel, with a dry run;
• which tags and confidentiality levels apply, especially for documents containing personal data;
• how a document's title is built — in NG DMS it is set with a template at intake.

In NG DMS an administrator creates types, attributes and registers on screen, without a developer — see Classification and metadata.

3. Paper preparation

Preparation is often the slowest part of the job, and everything after it depends on it:

• remove paper clips, staples and sticky notes; if a note carries information, stick it on a blank sheet and scan it as a separate page;
• flatten folded corners and tape torn pages;
• check the page order and pull out documents that do not belong in the stack;
• put an ASB label on the first page of every document and a separator sheet between documents;
• note which binder or box each stack came from, so the originals go back to the same place.

How labels are printed and how a separator splits a stack into documents is covered in ASB labels and separator sheets.

4. Scanning and a pilot

For text documents 300 DPI is the usual resolution, and the one at which NG DMS reads barcodes by default. Scan in colour when colour carries meaning — a stamp, a signature, highlighted text — and in greyscale when it does not. A scanner that scans one side only is no obstacle: odd pages go first, then the flipped stack with the even pages, and the system merges them into one document.

The route into the system depends on the installation. On-premises, the scanner drops files into a watched folder, the system picks a file up only once it stops changing, and subfolders become tags; “Process once” also brings in a folder that has already been scanned. In the cloud, the scanner sends to the organization's mailbox or over SFTP.

Before a large batch, scan one binder and take it all the way: from scan through OCR and description to a search by word. A pilot shows the real pace and the flaws in your rules while they are still cheap to fix.

5. OCR and the archival copy

A scan needs text. NG DMS produces the text of every page through OCR for Serbian in Cyrillic, Serbian in Latin script and English, then an archival copy in PDF/A-2b with a text layer; the original never changes. If the OCR service is temporarily down, the document is still taken in and its text follows later. Details are in OCR for Cyrillic and Latin script and PDF/A.

6. Classification and metadata

Nobody wants to describe thousands of scans by hand. NG DMS suggests the type, correspondent, tags and date from the organization's rules and a model that learns from your confirmed documents, while NG Copilot reads amounts, dates and numbers from the text and shows the source. The user confirms the suggestion at the validation station. A tag barcode and the scanner's subfolder add tags right at intake. When the system is not sure, the field stays empty — empty beats wrong.

7. Quality control

Check a sample from every box while the work is under way:

• the scan has as many pages as the paper;
• pages are legible and the right way up — a page that is upside down is rotated in the viewer, as a new version;
• every document has an ASB and a type, and required fields are filled in;
• the processing log shows no failed steps — every step records its outcome: done, skipped or failed, with the reason;
• duplicates were caught by their SHA-256 fingerprint before they became documents;
• a test search for a word on the scan finds the document, in both Latin and Cyrillic script.

8. Digitization record and integrity

For scanned paper NG DMS keeps a digitization record: who, when, where, with which device and procedure. Every file also gets a SHA-256 fingerprint at intake, a weekly integrity check compares the files with the record, and every change goes into a hash-chained audit trail. The mechanisms are summarised on the Compliance page.

What should you do with paper originals after scanning?

A scan is an electronic copy of a document, not an automatic replacement for the original. Whether the original is kept after scanning, and for how long, is set by the regulations, your list of record categories and, for archival material, the competent archive. Note too that technical integrity is not the same as legally trusted electronic preservation — the difference is explained in Trusted electronic preservation: what it is and what it is not.

In practice, originals go back into their boxes and binders after scanning, and the ASB number links them to the electronic copy. The physical archive in NG DMS records the location, box and binder of each original, and a loan is tracked with a loan record and a return date.

How do you get started?

The whole path from file to searchable document is described on the Capture page. To see it on your own scanner and your own documents, book a demo.