# How to Digitize Law Firm Paper Files Step by Step

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2024-04-14T05:15:04Z
- Updated: 2026-10-11T19:22:20Z
- Category: Converting scans
- Canonical: https://www.schemalaw.com/blog/how-to-digitize-law-firm-paper-files

> How to digitize law firm paper files: decide scope, set the scanner, name the files, run OCR on the backlog, check quality, and handle the paper.

To digitize law firm paper files, decide which files are worth scanning, scan them at settings OCR can read, name them by client and matter, run OCR so the text can be searched, check a sample against the paper, and only then decide what happens to the originals. The order matters more than the equipment.

A project is likely to stall when scanning starts before anyone has agreed on scope, names or what "done" means. The six sections below follow the order in which the decisions should be made.

## What should a firm decide before scanning anything?

Before scanning anything, a firm should decide which files are in scope, what each scanned file must be usable for, who does the work, and how it will know the result is good. Write the answers on one page. Every later step refers back to it.

### Which files

Do not scan everything. Sort the paper into groups first:

| Group | Usual decision |
| --- | --- |
| Open matters | Scan now, and scan new paper as it arrives |
| Closed matters still inside the firm's retention period | Scan if they are consulted often or storage is costly |
| Closed matters past the retention period | Review for destruction or return under firm policy; scanning them first is often wasted effort |
| Original documents held for clients, such as wills and deeds | Scan for reference and keep the paper |
| Firm administration: accounts, personnel, insurance | Follow the rules that apply to each record type |

The third row saves the most money. A box that could lawfully be destroyed or returned today should not be scanned, named and stored first.

### What the scans must do

"Searchable" and "editable" are different targets. If the goal is to find a closed file and read it, a searchable PDF is enough. If lawyers will reuse language from old agreements, they need Word exports of selected documents, not of every page in the archive. Decide per group.

### Who and when

Name one person who owns the project. Decide whether staff scan in quiet hours or an outside bureau takes the boxes. If a bureau is used, the files leave the building, so confidentiality terms and chain of custody need settling before the first box goes.

### What done means

Agree a test in advance, such as "any closed file can be found by client name and matter number, opened, and searched for a word, within two minutes". A test of that kind tells you whether the project worked. A page count does not.

## Scanner settings that make OCR work

Scanner settings that make OCR work are modest: 300 dots per inch, grayscale or color for anything other than clean black print, both sides of the sheet, and PDF output with light compression. Higher settings mostly produce bigger files. Lower ones cost you characters.

### Resolution

Use 300 dpi for ordinary typed pages. For a benchmark, the federal regulation on digitizing permanent paper records, [36 CFR 1236.50](https://www.law.cornell.edu/cfr/text/36/1236.50), requires image files at a minimum of 300 ppi for modern textual paper records. That rule binds federal agencies, not law firms, but it is a sound floor to copy. Go to 400 dpi for small print, such as footnotes, fine-print terms and old carbon copies. Fax resolution and the "draft" or "small file" presets on office copiers fall well below 300 and should be turned off.

### Color mode

| Mode | Use for | Watch out for |
| --- | --- | --- |
| Black and white | Clean laser-printed text on white paper | Faint text, pencil, highlighter and stamps can vanish or turn into black blocks |
| Grayscale | Most file contents: letters, pleadings, photocopies, faded pages | Larger files than black and white |
| Color | Pages where color carries meaning: highlighted exhibits, colored stamps, ink signatures, photographs, plans | The largest files |

When in doubt, choose grayscale. A black and white scan makes a hard decision about every dot, and a wrong decision cannot be undone later. Grayscale leaves that judgment to the OCR software.

### Both sides, and blank pages

Scan duplex. A missed reverse side is an easy loss in a digitization project, and nobody notices until the paper is gone. Blank-page removal keeps files tidy, but set it cautiously. A page with one faint handwritten line or a lone stamp can be judged blank and dropped.

### Preparation

Remove staples and clips, unfold corners, and flatten creases. Tape torn pages. Put small items, such as receipts and sticky notes, on a carrier sheet, and scan a sticky note twice: once in place, and once moved aside so the text under it shows. Scan fragile or bound originals on the flatbed, not through the feeder.

### File format and size

Save as PDF, one file per document, with the scanner's own OCR switched off if you plan to run OCR afterwards, so the file does not end up carrying two text layers. Avoid the highest compression settings, which smear small type. Watch the file size as well. Schema Law accepts PDF, PNG and JPEG uploads up to 20 MB, so a long color scan may need splitting into parts before upload.

Feed pages straight and upright where you can. For pages that arrive poor whatever you do, the [triage checklist for low quality PDF scans](https://www.schemalaw.com/blog/ocr-low-quality-pdf) covers what to try.

## How should scanned files be named and organized?

Scanned files should be named so that a stranger could find a document without opening it: client, matter, date, document type and a short description, in a fixed order. Mirror the paper file's structure in folders. Never rely on the names a scanner assigns.

A pattern that works for most firms:

```text
ClientNumber-MatterNumber_YYYY-MM-DD_Type_Short-description.pdf

01482-0007_2019-06-14_Corr_Letter-to-opposing-counsel.pdf
01482-0007_2019-07-02_Plead_Answer-and-counterclaim.pdf
01482-0007_2019-07-02_Plead_Answer-and-counterclaim_OCR.docx
```

The rules behind the pattern:

- **Numbers before names.** Client and matter numbers do not change spelling. Keep the leading zeros so files sort correctly.
- **Dates as year, month, day.** Files then sort in date order by name alone. Use the document's date, not the scan date.
- **A short, fixed list of types.** Corr, Plead, Disc, Agmt, Notes, Bill, or the divisions your paper files already use. Publish the list and do not let it grow.
- **No spaces or punctuation beyond hyphens and underscores.** Some systems mishandle them.
- **The same base name for every version.** The scan, the searchable PDF and any Word export share a name, with a suffix for what each is.

For folders, copy what staff already know: one folder per client, one per matter, then the same subdivisions the paper file had. A familiar structure needs no training.

Decide what counts as one document. One PDF per letter, pleading or agreement is far more useful than one PDF per box or per folder. A 600-page "Correspondence 2018" file can be searched but not sorted, cited or shared sensibly. The exception is a bound record that was always a single item, such as a transcript or a closing binder.

Give a last will and testament its own file, and keep the paper original. Name the scan so you can find the decedent, the date and the executor without opening it. A codicil stays with the will if they were fastened together. The searchable text lets you find a beneficiary or a bequest later, while the signed pages remain the record.

Keep an index as you go. A spreadsheet with one row per box is enough: box number, client and matter, date range, number of files, who scanned it, the date, and whether it passed checking. That index later becomes the record of what was scanned and what happened to the paper.

## Running OCR on a backlog of closed files

Running OCR on a backlog of closed files is a scheduling job. Count the pages, match them to the monthly capacity of your tool, start with the files most likely to be opened again, and track each box from scanned to recognized to checked.

### Count before you start

Count the sheets in three typical boxes, average them, and multiply by the number of boxes. Double the figure for the share that is two-sided. Suppose 40 boxes average 2,000 page images each: that is 80,000 pages.

Then compare with plan capacity. Schema Law's plans, as listed on the [pricing page](https://www.schemalaw.com/pricing), allow:

| Plan | Pages a month | Largest single document |
| --- | --- | --- |
| Solo | 1,000 | 250 pages |
| Team | 10,000 | 1,000 pages |
| Enterprise | 100,000 | No fixed page limit |

At 10,000 pages a month, the 80,000-page backlog takes eight months if nothing else uses the allowance. That may be acceptable for closed files. If it is not, the choice is a larger plan for the length of the project or a smaller scope. The post on [OCR cost per page](https://www.schemalaw.com/blog/ocr-cost-per-page) works through the price arithmetic.

### Order of work

1. **Open matters first.** They pay back at once.
2. **Recently closed matters next.** They are the likeliest to be reopened or requested.
3. **Old closed files last,** and only those inside scope.

### Pick the output per group

For archive files, export a searchable PDF: the page stays as scanned and gains text that can be searched. Export Word only for documents someone will actually reuse. The guide to [making a scanned PDF searchable](https://www.schemalaw.com/blog/how-to-make-a-scanned-pdf-searchable) explains that export and how to test it.

### Keep the batch in order

Upload in the same units you named: one document per file, a box at a time. Mark each box in the index when its uploads are done and again when its exports are saved to the firm's own storage. On every plan you can choose several files at once, and they are read one after another. Each account belongs to one person, and no other account can see its documents, so if several people share the scanning, agree who uploads which boxes and gather the exports in the firm's own storage. The post on [batch OCR for discovery documents](https://www.schemalaw.com/blog/batch-ocr-discovery-documents) covers splitting and tracking large sets.

### Do not leave the archive in the tool

A conversion tool is not the firm's records system. Download the exports, file them with the scans, and then decide when the uploaded copies should go. Schema Law deletes nothing automatically: a document stays until you delete it, so make deletion a column in the index. The post on [whether OCR software stores your documents](https://www.schemalaw.com/blog/does-ocr-software-store-my-documents) explains what to ask of any tool.

## Quality control for a digitization project

Quality control for a digitization project means checking a sample of every batch against the paper before the paper moves on. Check completeness first, then legibility, then the text. Record each check. A project with no checking record cannot show that its scans are reliable.

### Completeness

This is the check that matters most, because a missing page cannot be fixed once the original is gone.

- Compare the page count of each PDF with the sheets in the folder, allowing for two-sided pages.
- Look at the first and last page of every document.
- Fan through the paper for small or odd-sized items the feeder may have skipped.
- Look for pages that fed two at a time. A sudden jump in page numbering is the sign.

### Legibility

Open a sample of pages at full size. Can you read the smallest print, the faintest stamp and any handwriting as well as you can on the paper? Are pages upright and uncropped? Did highlighter turn into a black bar?

### Text

Search for a word you can see on the page. Then read a few lines of the recognized text against the image, concentrating on names, dates, amounts and section numbers. The post on OCR errors has a full method for [proofreading OCR output against the original](https://www.schemalaw.com/blog/why-is-ocr-output-full-of-errors#proofreading-ocr-output-against-the-original).

### How much to sample

A workable scheme for a small firm:

| Stage of the project | What to check |
| --- | --- |
| First three boxes | Every document for completeness; one page in ten for legibility and text |
| Steady state | Every document's page count; five documents a box in full |
| After any change of operator, scanner or settings | Back to the first-stage level for one box |
| Any box that fails | Rescan the failed items, then check the whole box |

Write the result in the index: who checked, when, what was sampled and what was rescanned. Keep failed scans out of the archive, or they will be found later and trusted.

Set one rule that nobody may break: no box leaves for storage or destruction until its row in the index shows a passed check.

## What happens to the paper afterwards

What happens to the paper afterwards is a legal and ethical question first and a storage question second. Depending on the document, the paper is kept, returned to the client, or destroyed securely. The rules on client files and retention periods vary by jurisdiction, so check yours before destroying anything.

This post describes how to handle the documents. It is not legal advice on what your bar requires. Use it to frame the questions for whoever is responsible for the firm's records policy.

### Sort the paper again

After scanning and checking, each item falls into one of four groups:

| Group | Examples | Usual handling |
| --- | --- | --- |
| Originals with legal effect of their own | Signed wills, original deeds, promissory notes, share certificates, documents under seal | Keep the paper, or return it to the client against a receipt. The scan is a reference copy |
| Client property | Documents and items the client supplied | Return or keep, as the client and your rules direct |
| Files subject to a hold | Anything relevant to pending or expected litigation, an audit or an investigation | Keep until the hold is released |
| Ordinary file copies | Photocopies, printouts of emails, drafts, duplicates | Candidates for secure destruction once the scan has passed checking and policy allows |

### Questions to settle before destruction

1. What does your jurisdiction say about keeping, returning and destroying client files, and for how long?
2. Do your engagement letters or file-closing letters tell clients what happens to their file?
3. Does your malpractice insurer have a position on scanned files replacing paper?
4. Is any matter under a litigation hold?
5. Is the digital copy backed up, and has a restore been tested?

### Destroying what can go

Use cross-cut shredding or a destruction service that gives a certificate. Record the box, the date, the method and who authorized it in the same index that tracked the scanning. That one spreadsheet then shows, for any file, when it was scanned, who checked it, and what became of the paper.

Consider holding scanned paper for a set waiting period before destruction, so that gaps found in daily use can still be fixed. Whatever period you pick, write it down and apply it to every box.

A digitization project ends when the index is complete, not when the last box is scanned. To try the OCR step on one of your own documents before planning the rest, the [three steps on the homepage](https://www.schemalaw.com/#how-it-works) show how an upload becomes an export.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)