The Rules of a Naming Convention That Holds Up
A naming convention is only worth adopting if it survives contact with a real folder full of hundreds of files and more than one person adding to it. Six rules do most of the work. Each one is small on its own, and each one is easy to see in a before and after.
Be descriptive. The filename should say what the file is without anyone opening it. Scan_0042.pdf tells you nothing; 2026-03-15_AcmeCorp_Invoice_INV-2847.pdf tells you the date, the vendor, the document type, and the invoice number at a glance. A descriptive name is the difference between reading a folder and searching one.
Use ISO dates, YYYY-MM-DD. Written this way, dates sort chronologically in any file browser, and there is no ambiguity about whether 03-04 means March 4 or April 3. Before: Invoice_Mar15.pdf. After: 2026-03-15_Invoice.pdf. The ISO date is the single highest-leverage rule because it makes the whole folder sortable for free.
Put the most important element first. Files sort left to right, so the field you search by should lead. A team that pulls records by date leads with the date; a team that pulls by client leads with the client. Before: Invoice_AcmeCorp_2026-03-15.pdf scattered among vendors. After: 2026-03-15_AcmeCorp_Invoice.pdf, where every file lines up by the thing you actually look for.
Be consistent. Same field order, same separators, every single time. Consistency is what lets a filename be scanned by eye and by software. Before: a folder mixing Acme-Invoice-Mar.pdf, invoice_acme_0315.pdf, and ACME INVOICE.pdf. After: three files that share one shape and sort together.
Avoid spaces and special characters. Spaces break links, scripts, and command-line tools, and characters like slashes or colons are illegal in many systems. Use hyphens or underscores instead. Before: Acme Corp Invoice (final).pdf. After: 2026-03-15_AcmeCorp_Invoice_final.pdf.
Include a version or status where it matters. For anything that goes through revisions or approvals, a trailing v2 or Signed removes the guesswork. Before: Contract_final_FINAL_use-this.pdf. After: 2025-11-04_Acme-Meridian_MSA_Signed.pdf, where the status is a field, not a plea.
Building Your Template: Pick the Fields, Lock the Order
The rules above become a convention when they are frozen into a template. A template is just an ordered list of fields with fixed separators, written once so nobody has to reinvent it per file. The building blocks are almost always the same handful of fields: the date, the party or client, the document type, and a reference such as an invoice number, matter code, or account.
Pick the fields that matter for a given document type, decide their order, and choose one separator scheme. A common and readable choice is an underscore between fields and a hyphen inside a multi-word field, so 2025-11-04_Acme-Meridian_MSA_Signed.pdf reads cleanly at a glance. What matters is not which separators you pick but that you pick once and never vary them.
Write the template as a literal field-order string, for example {date}_{party}_{doctype}_{reference}, and keep a short table of one template per document type. That table is the entire convention. The next section is exactly that table, a copy-ready starting point you can adapt to your own document types.
| Document type | Template (field order) | Example |
|---|
| Invoice | {date}_{vendor}_{doctype}_{number} | 2026-03-15_AcmeCorp_Invoice_INV-2847.pdf |
| Contract | {date}_{parties}_{doctype}_{status} | 2025-11-04_Acme-Meridian_MSA_Signed.pdf |
| Statement | {date}_{institution}_{doctype}_{account} | 2026-01-31_Chase_Statement_1234.pdf |
| Report | {date}_{project}_{doctype}_{version} | 2026-02-10_Q1_Report_v2.pdf |
The Enforcement Problem: Conventions Drift
Writing the template down is the easy part. The hard part is that a convention only works if everyone applies it identically, and manual naming drifts the instant a second person is involved. One person types Invoice, another types INV, a third abbreviates the vendor differently, and a fourth forgets the date because the file was in a hurry. None of them is wrong on purpose. They are just human, and human naming is inconsistent by default.
Drift is not a discipline problem you can train away with a wiki page. The moment a folder receives files from more than one source, an email attachment here, a scan there, an export from an accounting tool, the names diverge. A convention maintained by willpower degrades a little every week until the folder is back to being a pile you have to search instead of read.
The before and after below shows what drift looks like on four files that should have been identical in shape, and what they look like once the same template is applied to all of them.
| Before (drifted) | After (convention applied) |
|---|
| INV acme 3-15.pdf | 2026-03-15_AcmeCorp_Invoice_INV-2847.pdf |
| Contract FINAL v2.pdf | 2025-11-04_Acme-Meridian_MSA_Signed.pdf |
| chase statement.pdf | 2026-01-31_Chase_Statement_1234.pdf |
| Q1 report (2).pdf | 2026-02-10_Q1_Report_v2.pdf |
Enforcing the Convention Automatically
The reliable fix is to stop making the name something a person types and make it something the tool generates. If the filename is produced from the document's own content every time, there is no gap for drift to creep into: the same input always yields the same name, no matter who dropped the file in the folder.
Unlike rule-based renamers that only see filenames, renamer.ai reads the actual document content (OCR + AI vision) to generate descriptive names automatically. It opens each file, identifies the fields the template calls for, the date, the vendor or parties, the document type, the reference number, and writes them out in the exact order and separators the convention specifies. A folder of forty scans becomes forty files that follow one pattern, without anyone hand-typing a single name.
Because the fields come from the content rather than from memory, the convention holds across a whole team. Renamer.ai applies the same template to every file it processes, so the person who would have typed INV and the person who would have typed Invoice both end up with the identical, correct name. The convention stops being a rule people have to remember and becomes a property of the folder.
Making the Convention Work on Scans and OCR
A naming convention is only enforceable on scanned documents if a tool can actually read them, which is what OCR provides. A scanner names a file by the minute it ran, Scan_20260315_0941.pdf, because it captured an image and never read the words on the page. Nothing about the vendor, the date on the invoice, or the contract type is available to name from, so the template has nothing to fill.
Optical character recognition is what turns that image back into text a convention can use. Once the words on the page are readable, the template's fields become fillable: the invoice date printed on the document, the counterparty named in a contract, the account number on a statement. That is what makes a convention enforceable on a stack of scans rather than only on files that arrived as clean digital PDFs.
For a deeper walkthrough of applying names to image-only documents, see the hub guide on how to rename scanned PDF files, which covers the scanned-document case in detail.