The Complete Guide to Bates Numbering and Legal Document Stamping

Document stamping is one of those tasks that nobody goes to law school for and everybody ends up doing. Somewhere between the discovery request and the production deadline, a pile of PDFs has to be numbered, labeled, watermarked, and checked, and the quality of that work quietly shapes how the rest of the case goes. A clean Bates range makes depositions faster, motions easier to draft, and privilege disputes easier to resolve. A sloppy one creates duplicate numbers, missing pages, and awkward letters to opposing counsel explaining why the production has to be re-served. Most of the pain is avoidable, but only if the team treats stamping as a process with rules rather than a chore to rush through the night before the deadline.

This guide is our attempt to lay out that process end to end. We build stamping software for a living, so we spend a lot of time talking to paralegals, litigation support staff, and solo practitioners about what actually goes wrong. The patterns repeat: prefixes chosen without thinking about the next production, scanned files that render stamps in the wrong place, originals overwritten by accident, and numbering verified by eyeballing a few pages instead of checking the whole set. The seven themes below cover the fundamentals of Bates numbering, prefix and range design, watermarks and exhibit labels, file preparation, batch execution, verification, and building a repeatable workflow. Each section links to the deeper articles on this site where you can go further on a specific problem.

Bates Numbering Fundamentals: What It Is and Why It Persists

Bates numbering is the practice of stamping a unique identifier on every page of a document set so that any page can be cited, located, and matched without ambiguity. The identifier usually combines a prefix, often an abbreviation of the producing party or matter, with a zero-padded sequential number, producing labels like ACME000001. The name comes from a hand-operated numbering machine that predates modern litigation by generations, but the underlying need is much older and much simpler: when thousands of pages change hands, people need a shared language for pointing at one of them. Page 14 of the third binder means nothing to opposing counsel. ACME000342 means exactly one page, and it means that same page for as long as the case exists.

That permanence is the reason the practice survived the move from paper to PDF. In litigation, a Bates number gets written into deposition transcripts, privilege logs, exhibit lists, motions, and court orders. Once a number is out in the world it cannot be reassigned without breaking every reference that points at it, which is why experienced teams treat the first production as a commitment rather than a draft. The number also carries a quiet audit function. A gap in the sequence, a duplicate, or a page that appears under two different numbers is immediately visible to anyone reviewing the set, which makes deliberate or accidental omissions harder to hide and easier to discuss with the other side in concrete terms.

It helps to separate the concept from the mechanics. Conceptually, Bates numbering is about uniqueness, sequence, and stability. Mechanically, it is about placing readable text on a page at a consistent position, in a consistent font, without covering content, and doing that identically across a set that might include native PDFs, scanned images, spreadsheets printed to PDF, and long email threads. The mechanics are where software matters and where most mistakes happen, but the concept is what determines whether a scheme will hold up two years into a case. If you are new to the topic, start with our explainer on what Bates numbering is and why law firms rely on it, then read about prefix conventions before you stamp anything.

Designing Prefixes and Ranges That Survive the Whole Case

The single most consequential decision in a stamping workflow is made before a single page is stamped: what prefix and what number format will the matter use? A prefix should identify the producing party clearly, stay unique within the matter, and remain readable when abbreviated in a transcript or a spreadsheet. Short alphabetic prefixes tend to work best. Long descriptive prefixes look informative on day one and become clutter by the third production, and prefixes that resemble each other, such as two co-defendants whose names start with the same three letters, invite confusion in exactly the moments when precision matters most. A firm-wide convention that everyone follows without thinking is worth more than a clever scheme that only its author understands.

Number width matters almost as much as the prefix. Zero-padding to a fixed number of digits keeps labels sorting correctly in file systems, review platforms, and spreadsheets, and it removes the visual ambiguity between ACME1 and ACME10000. The right width depends on realistic volume, not on the first batch. A matter that starts with a few hundred pages can easily reach tens of thousands after a second custodian is added, and running out of digits midway forces either an ugly format change or a renumbering that breaks every existing citation. Most teams settle on a width that comfortably exceeds any volume they expect, and then they never touch it again for the life of the matter, no matter how small the first production turns out to be.

The prefix and width are only half of the design. The other half is a rule for how ranges are assigned across productions. Each production should pick up exactly where the last one ended, with no overlap and no unexplained gaps, and someone has to own the ledger that records the last number used. That sounds obvious, but in practice the ledger lives in a paralegal's head, or in a spreadsheet that two people edit at once, and the next production starts at a number that was already used. Our article on keeping numbering consistent across multiple productions walks through a simple ledger discipline, and the prefix conventions guide covers the naming side in more depth, including how to handle supplemental and rolling productions without reinventing the scheme each time.

Watermarks, Confidentiality Legends, and Exhibit Labels

Bates numbers identify pages; other stamps describe them. The most common companions are confidentiality legends applied under a protective order, privilege markings on documents withheld or produced with redactions, and exhibit labels used at depositions, hearings, and trial. Each of these has different rules about placement, wording, and timing, and each interacts with the Bates number in ways that are easy to get wrong. A confidentiality legend that overlaps the Bates number, or an exhibit sticker that covers the only copy of a signature, creates an argument that could have been avoided with a few minutes of planning and a quick preview of the stamped page before the batch runs.

Confidentiality designations are usually governed by the protective order in the case, which often specifies the exact wording, sometimes the tiers (for example a standard confidential tier and a more restrictive attorneys-only tier), and occasionally the placement. The practical challenge is deciding which documents get which legend and applying that decision consistently across the batch. Blanket designation of everything as confidential is tempting and fast, but many protective orders discourage it and opposing counsel can challenge over-designation. A better approach is to sort documents into designation groups during review, then stamp each group as its own batch with the right legend, so the designation decision is made once and applied mechanically. Our article on when to apply confidential and privileged watermarks goes deeper on timing and wording.

Exhibit labels serve a different purpose and follow different conventions. They are typically applied to a copy of the document, not to the production set, and they identify the exhibit number, the case, and often the date or the witness. The traditional sticker in the corner has a digital equivalent that can be placed with the same care: away from content, consistent across the exhibit binder, and sized to remain legible when the page is printed or displayed on a courtroom screen. Because exhibits are often assembled from already numbered productions, the label should never obscure the Bates number, since the transcript will reference both. The exhibit labeling best practices article covers positioning, numbering sequences for multi-party depositions, and how to prepare an exhibit set the night before a hearing without surprises.

Preparing the Files: Originals, Scans, and Text Layers

Stamping is a transformation, and every transformation should start from a protected source. The most damaging mistake we see is not a wrong number but a lost original: a team stamps the only copy of a client's files in place, discovers a numbering error, and now has no clean version to restamp. The fix is a habit, not a tool. Originals go into a read-only folder with a manifest that records file names, hashes, and page counts, and every stamping run writes to a separate output folder. If a run goes wrong, you delete the output and try again. If you ever need to prove what the client actually handed over, the untouched originals are there. Our article on preserving original documents explains how to structure that folder layout and why hashes matter.

The second preparation issue is the nature of the PDF itself. Native PDFs generated from word processors or email exports have a text layer, predictable page sizes, and clean margins, so stamps land where you expect. Scanned PDFs are pictures of paper. They may be skewed, have irregular page sizes, contain mixed portrait and landscape pages, and lack any text layer at all. A stamp placed relative to the page edge can end up sitting on top of a scanned header or falling outside the visible area entirely if the scan has a large black border. Stamping a scanned set without previewing a sample of the odd pages is how teams end up with numbers printed across the middle of a signature block.

Text layers matter for more than placement. Without optical character recognition, a scanned PDF cannot be searched, and some review tools cannot extract the stamped Bates number as metadata from an image-only page. That does not mean you must OCR everything before stamping, but it does mean deciding deliberately. Running OCR first produces a searchable, stampable file; stamping first and then running OCR can cause the recognition engine to pick up the stamp text as part of the document content. Our guide to stamping scanned image PDFs walks through how to detect image-only files, how to normalize page sizes, and when it is worth cleaning up scans before the batch runs.

Running the Batch: Thousands of Pages Without Losing a Day

Once the scheme is designed and the files are prepared, the batch itself should be the boring part. The goal is a run that is fast, deterministic, and easy to repeat. Fast because a small team cannot afford to babysit a progress bar for hours; deterministic because running the same input with the same settings must produce the same output, so a re-run after a fix does not introduce new differences; and repeatable because most matters involve several productions over months, and each one should follow the same recipe. A workflow that requires a specific person to remember a specific sequence of clicks is not a workflow, it is a risk that happens to be working for now.

The practical structure of a good batch run is straightforward. Sort the documents into the order the production will use, usually by custodian and then by date or by the order they were collected. Confirm the starting number from the ledger. Set the stamp position, font, and size once, and apply them across the entire set rather than per file. Stamp into an output folder, never in place. Then produce a load file or a simple CSV that maps each source file to its beginning and ending Bates number, because that map is what the review platform, the privilege log, and the production letter will all need. Our article on batch-stamping thousands of pages breaks down each of these steps and explains where the time actually goes.

Two things reliably slow down large batches: files that do not open cleanly, and pages that do not fit the expected layout. Corrupt PDFs, password-protected files, and documents with embedded forms or annotations can each halt a run or produce silently wrong output. The defensive move is to run a validation pass before stamping, flag anything that cannot be opened or has zero pages, and handle those files separately rather than letting one bad file stop a thousand good ones. The same logic applies to landscape pages, oversized engineering drawings, and mixed page sizes. Treat the exceptions as their own small batch with their own settings, and the main run becomes predictable again. That discipline pairs naturally with the original-preservation habits described earlier, since a separate exception folder is much easier to manage when the originals are already isolated.

Verification: Proving the Numbers Are Right Before Anyone Else Sees Them

Every stamping workflow should end with a verification step that is independent of the stamping step. That independence is the point. If the same tool that applied the numbers is also the one reporting that the numbers are correct, you have not really checked anything. Verification means opening the output set and confirming, page by page or at least range by range, that the first number matches the ledger, that the last number equals the first number plus the page count minus one, that no number appears twice, that no number is missing, and that the stamp is legible and correctly positioned on a representative sample of every document type in the set.

The arithmetic checks are the easy part and can be automated: extract the stamped text from each page, sort it, and look for gaps and duplicates. The harder checks are visual. A number that is technically present but printed in white on a white margin, or sitting on top of a table row, or rotated with a landscape page so it reads sideways, will pass the text extraction and fail in front of a judge. The compromise most teams reach is full automated arithmetic verification plus a human review of a structured sample, weighted toward the documents most likely to be cited: contracts, key emails, and anything with a signature or handwritten annotation. Our article on verifying numbering before production includes a checklist you can adapt to your own matters.

Verification also closes the loop with the ledger. After the run passes, the last number used gets recorded, along with the production volume label, the date, and a hash of the output set. That record is what makes the next production start in the right place and what lets you answer, months later, exactly which version of which document went out under which number. Teams that skip this step do not usually notice until a supplemental production overlaps an earlier range, and by then the fix is expensive and embarrassing. The consistency article covers how to structure that ledger so that multiple people can maintain it without stepping on each other.

Building a Repeatable Document Ops Process for a Small Firm

Large firms have litigation support departments and vendors to absorb this work. Small firms and solo practitioners have a paralegal, an associate, and a deadline. The good news is that the process scales down well if it is written down. A one-page standard operating procedure that records the firm's prefix convention, the folder structure for originals and output, the default stamp position and font, the ledger location, and the verification checklist will save more time than any single software feature. It also makes the work transferable. When the person who usually handles productions is on vacation, someone else can follow the document instead of guessing, and the production goes out looking exactly like the last one.

Tooling decisions come after the process is defined, not before. Whether a firm uses a desktop PDF editor, a command-line script, a dedicated stamping application, or a review platform's built-in numbering, the questions are the same: can it stamp a folder of mixed PDFs in one run, can it write to a separate output location, can it produce a Bates map, can it handle scanned pages without manual repositioning, and can someone verify its output independently? Tools that answer yes to all five can support a defensible process. Tools that require per-file manual work will be fine for a twenty-page exhibit set and painful for a five-thousand-page production. We are obviously not neutral on this, since we build stamping software, but the criteria hold regardless of what you choose.

The final ingredient is review. After each production, spend fifteen minutes noting what went wrong and what took longer than expected, then update the procedure. Did a prefix collide with another party's? Add a check to the ledger. Did scanned pages need manual fixes? Add a preview step for image-only files. Did verification miss a sideways stamp? Add landscape pages to the visual sample. Over a few matters, this turns a stressful scramble into a routine that a new hire can run in their first week. The fundamentals article and the prefix conventions guide are good onboarding reading for anyone joining that routine, and the batch-stamping article is the natural next step once they understand the why behind each rule.

More guides on this topic

Further reading from the PdfStampr blog, each answering one specific question in depth.

Bates numbering and document stamping are not glamorous, but they are load-bearing. Every citation, exhibit, and privilege log entry in a case rests on the assumption that the numbers are unique, stable, and legible. Getting there is mostly a matter of sequence: design the prefix and width before you start, protect the originals, understand what kind of PDFs you have, run the batch into a separate folder with one set of settings, verify independently, and record the result in a ledger someone owns. None of those steps is difficult on its own. The failures we see come from skipping one of them under deadline pressure. If you take one thing from this guide, make it the habit of writing the process down so that the next production, and the person running it, starts from a known good state rather than from memory.