How to Digitise a Paper Archive Without Losing Your Mind

Digitisation projects rarely fail at the scanner. They fail six months in, when there are forty thousand images called scan0001.pdf, nobody knows which boxes are done, and the paper has been thrown away. A project that survives has four parts decided before the first page is scanned: what to scan and in what order, how files are named and described, how quality is checked, and what happens to the originals.

1. Decide what to scan, and what not to

Not everything needs digitising. Start with an inventory at box or file level and sort it:

• Scan first: records used often, records at risk (fading, damp, fragile), and records needed for ongoing work.
• Scan later: records kept for completeness but rarely consulted.
• Do not scan: records past their retention date. Destroy them properly instead. Scanning them only creates a digital copy you will also have to destroy; see proving you deleted what you said you would.

2. Fix the naming and description scheme

Decide the file naming rule before anyone scans, and make it predictable: collection, box, file, item, in that order, with leading zeros so files sort correctly. Our guide to file naming conventions people actually follow has patterns that work. Then decide the minimum description for each file or item: title, date range, creator, and access restrictions. Describing at file level is usually enough; item-level description for everything is how projects stall.

3. Choose formats and settings once

• A preservation master at a resolution that captures the smallest text and any colour that matters, in an open, well-documented format.
• A searchable PDF or similar access copy, with text recognition, for everyday use.
• Settings written down, so the tenth batch matches the first.

See file formats that will still open in twenty years for the choices.

4. Track progress at box level

Keep a simple tracker with one row per box or file: status (not started, scanning, checked, described, done), who did it, dates, number of images, and notes about problems. This is the document that tells you where you are when the person running the project goes on leave.

5. Check quality as you go

1. Every batch: count the images against the pages.
2. A sample from each batch: is it legible, complete, correctly oriented, with no fingers or cut-off edges?
3. Text recognition spot-checks on anything people will search.
4. Record the check and who did it.

Errors found in week one are fixed in minutes. Errors found in month six mean rescanning boxes that may already be gone.

6. Decide what happens to the paper

This is a records decision, not a space decision. Some originals must be kept regardless of scanning: records whose legal value depends on the original, items of historical or artefactual value, anything with original signatures where your jurisdiction requires them. Others can be destroyed after scanning and checking, if your policy and any applicable law allow it. Write down the rule, apply it per series, and record the destruction.

7. Store and protect the result

Scans of personnel files or client records are just as sensitive as the paper, and easier to copy. Keep them in a system with access controls, at least two copies in different places, and a periodic check that files still open. A digitised archive that nobody can find their way around is a new problem; one that is described, named and checked is the reason you started.