Split Before You Scan: Ending the Single-PDF Habit in UK Practices

A scanner with an automatic document feeder invites volume. Load the stack, press start, and a single PDF appears on the network. For the person standing at the machine, the job is finished. For the fee earner, the compliance officer or the client waiting for a document set, it has only just begun.

The one-big-PDF habit is common in small UK practices because it is fast at the point of capture. The cost arrives later, when someone has to find page 47 of a 220-page file, redact a third-party name that should never have been in the same document, or respond to a subject access request without over-disclosing material from another matter. Batch scanning is not the problem. Scanning without boundaries is.

Why the single-file default survives

Most multifunction devices and desktop scanners are configured to produce one output file per scan job. That default suits a single letter. It does not suit a morning’s post, a completion bundle or a disclosure set. Operators often work under time pressure, and separating documents before scanning feels like an extra step. The result is a digital pile that looks orderly because it has a filename, but behaves like a shoebox.

The habit also survives because the immediate feedback is positive. The scan completes, the file saves and the paper goes into the recycling. Nothing visibly fails. The failure is deferred to retrieval, review or disclosure, where it is more expensive to fix.

Retrieval: a PDF is not a folder

Case management systems, document management systems and shared drives expect documents to sit at matter level. A single PDF containing correspondence for three clients, a bank statement for a fourth and a court notice for a fifth cannot be filed correctly without manual splitting. That work is usually done by a junior fee earner or an administrator who did not create the scan and does not know where one document ends and another begins.

Search makes the problem worse rather than better. A text search inside a monolithic PDF returns a page number, not a matter reference. The reviewer then has to establish context from the page itself, which may be a continuation sheet with no client name. Version control also suffers: if one page is rescanned or replaced, the whole PDF may be reissued, creating duplicate copies with near-identical names.

Redaction and disclosure exposure

A single PDF is a single disclosure unit. When a subject access request, litigation disclosure exercise or regulatory query arrives, every page in that file must be reviewed in the same context. If the file mixes matters, the review must separate them before redaction can begin. That increases the chance of over-disclosure, missed redactions and inconsistent treatment of third-party information.

OCR adds a further consideration. A searchable text layer is valuable, but it can also preserve text that appears to have been visually removed if redaction is applied incorrectly. Proper redaction removes the underlying text, not just the visible image. A batch that has been split by matter makes that review more controlled because each PDF has a clear boundary and a known purpose.

UK GDPR principles of data minimisation and storage limitation apply to scanned records as much as to any other format. Keeping unrelated matters in one file makes retention harder to apply consistently. A file that should be kept for one matter may end up retained because it also contains material for another.

Separator sheets as the first control

A separator sheet is a page placed between documents before scanning. It can be a plain coloured sheet, a pre-printed sheet with a matter reference, or a barcoded sheet that the scanner recognises. The scanner profile is then set to split the output at each separator. This turns a single physical stack into a series of digital documents without manual intervention after the scan.

For a small firm, the simplest reliable method is a coloured separator sheet with a handwritten or printed matter number and document type. Barcoded separators are faster at volume, but they require a consistent scanning profile and a supply of sheets at the machine. Whichever method is used, the separator must be removed from the final PDF. A visual check should confirm that no separator page has been captured as a document.

Sorting before scanning is essential. Post should be opened, date-stamped and grouped by matter before it reaches the feeder. A file should be checked for loose documents that belong elsewhere. The scan operator should not be asked to guess matter boundaries from content alone.

Matter-level splitting and naming

Each output PDF should carry a filename that identifies the client or matter, the document type and the scan date. A consistent convention such as client-matter-doctype-2026-11-13 is more useful than a scanner-generated number. Avoid spaces and special characters that cause problems across systems. Keep the convention short enough to type and long enough to distinguish similar matters.

Where a matter has several document types, split them into separate PDFs rather than one combined file. Correspondence, pleadings, financial records and identification documents serve different purposes and may have different retention rules. A per-matter split is the minimum; a per-document-type split is better where volume allows.

OCR and the visual QA pass

OCR should run as part of the scan profile, not as an afterthought. Check a sample page from each output PDF to confirm that the text layer is present and readable. Search for a known term, such as the client surname or a date, to confirm that OCR has worked. Handwritten annotations will not usually be captured, so note them separately if they matter.

A quick visual quality check takes less than a minute per batch. Review the thumbnails for missing pages, skew, double feeds and blank pages. Compare the page count in the PDF with the physical batch count, allowing for separators. Check that page orientation is correct. Confirm that the separator sheets have not been included. If a page is illegible, rescan it before the batch leaves the room.

Handover to case management

Once the PDFs are split, named and checked, import them into the matter file or case management system. Apply the correct retention rule at the point of import rather than relying on a later cleanup. Do not email a large batch as a single attachment; use a secure link or the firm’s approved transfer method. Record the scan date, the operator and the batch reference so that the digital file can be traced back to the physical documents if needed.

A short operational rule set

• Sort post and files by matter before scanning.
• Use separator sheets and a scanner profile that splits on them.
• Produce one PDF per matter, and separate by document type where practical.
• Apply a consistent filename convention that includes the scan date.
• Verify OCR on a sample page and search for a known term.
• Run a thumbnail check for missing pages, skew and separator capture.
• Import to the correct matter and apply retention at the point of filing.

The aim is not to slow down scanning. It is to keep the speed at the machine from becoming a delay at the desk. A batch that is split correctly at source takes seconds longer to prepare and minutes less to retrieve, review and disclose. For UK firms handling volume, that is the difference between a digital file and a digital pile.