Open the program tree of any marketing automation instance that's more than two years old and you can read its history like tree rings: three different naming styles, half-finished folder reorganizations, a program called Copy of Copy of Q3 Webinar FINAL v2. Every one of those layers marks the moment an admin left and the next one couldn't decode the system they inherited.
A naming convention isn't tidiness. It's the difference between a stack one person understands and a stack anyone can operate. It's what makes reporting filterable, cloning safe, and handovers survivable. Here's a schema that holds up at scale — the examples use Marketo, but the logic transfers to any platform.
The one rule that matters: name for the reader, not the writer
Whoever creates a program knows what it is. The name exists for everyone else: the analyst filtering a smart list six months later, the new hire trying to find the active nurture, the auditor reconstructing what ran last quarter. Every element in a name should answer a question a future reader will have — when did this run, what kind of thing is it, what's it about, where does it live.
That principle settles most naming debates before they start. If a token doesn't help someone find, filter, or trust the asset later, it doesn't belong in the name.
A program naming schema that scales
A structure that has survived contact with real instances:
YYYY-MM | Type | Region | Descriptor
For example: 2026-09 | WBN | EMEA | Lifecycle Reporting Deep Dive or 2026-Q3 | NUR | GLOBAL | Post-Trial Onboarding.
Why each token earns its place:
Date first, ISO order. YYYY-MM sorts chronologically in every alphabetical list, which is how platforms display things. Use the launch or start date; for evergreen programs, the quarter or year of creation. The date also gives you instant archaeology — anything dated three years ago is a candidate for review.
Type as a fixed abbreviation. A short controlled vocabulary — WBN (webinar), NUR (nurture), EML (standalone email), EVT (event), CON (content/gated asset), OPS (operational), TRIAL, PPC — makes program type filterable in reports and unmistakable at a glance. The vocabulary matters less than its enforcement: publish the list, keep it under fifteen entries, and reject inventions.
Region or business unit if your instance serves more than one. Skip it if you're single-market — empty tokens are noise.
A human-readable descriptor at the end, because after all the structure, someone still has to recognize the thing. Keep it short and specific; "Deep Dive Webinar" is worse than the webinar's actual topic.
Use a consistent delimiter throughout — pipes, dashes, whatever your platform displays cleanly — and never mix them.
Assets inherit from their program
Inside a program, local assets don't need to repeat the full program name — the platform already shows the hierarchy. They need a type prefix and a role: EM-01 Invitation, EM-02 Reminder, LP-01 Registration, LP-02 Thank You, FRM-01 Registration Form, SL-Members for the smart list of members.
Two habits pay off disproportionately. Number sequential assets (EM-01, EM-02) so send order is visible without opening anything. And name smart campaigns by behavior, not vaguely: TR-Send Reminder (Non-Responders) tells the reader it's a trigger and what it does; Flow 3 tells them nothing and dares them to click into it.
Folder structure: shallow, dated, and boring
Folder trees fail in two directions — too flat (everything in one folder, unfindable) and too deep (five levels of nesting nobody maintains). Three levels is almost always enough:
Top level by function: Marketing Programs, Operational, Templates & Cloning Sources, Archive. Second level by year. Third level by type or by quarter, depending on volume. That's it. The folder answers "where does this live"; the program name answers everything else, so the two systems don't need to duplicate each other.
Two folders deserve special respect. Templates & Cloning Sources holds the blessed, pre-tokenized program shells everyone clones from — this is where a naming convention actually gets enforced, because a correctly named template produces correctly named clones. Archive is where completed programs go on a schedule, not "eventually"; an archive folder that nobody moves things into is just a label.
Operational assets: the naming that prevents outages
The programs that route leads, sync data, manage scoring, and process subscriptions deserve the strictest naming in the instance, because they're the ones where ambiguity causes incidents. Prefix them unmistakably — OPS-Scoring | Behavioral, OPS-Routing | EMEA Assignment, OPS-Data | Country Normalization — and keep them in a folder with restricted edit access if your platform allows it.
When we audit an instance, unlabeled operational campaigns are the single most common source of "nobody knows what this does, and nobody dares turn it off." A prefix costs four characters and prevents that entire category of fear.
Governance: the part everyone skips
A convention that lives in one person's head is a convention for one admin generation. Three things make it durable: a one-page reference doc (the token vocabulary, the schema, five examples) linked where people actually work; enforcement at the cloning source, since most programs are clones and clones inherit names; and a quarterly ten-minute sweep for drift — new inventions in the type vocabulary, undated programs, things named test in production folders.
Retrofitting an old instance? Don't attempt a big-bang rename — reporting references, sync mappings, and muscle memory all break. Apply the convention to everything new starting today, rename operational programs carefully with a change log, and let the archive absorb the old chaos on schedule.
What good looks like
The test of a naming convention is a stranger test: could a competent marketing ops person who has never seen your instance find the active EMEA nurture, identify last quarter's webinar programs, and tell operational campaigns from marketing ones — in under five minutes, without asking anyone? If yes, your instance survives its next admin transition. If no, the knowledge lives in someone's head, and heads leave.
Instance structure and automation hygiene are part of what we review in a stack audit — a 45-minute walkthrough that ends with a prioritized fix list, yours to keep. For ongoing operations — templates, governance, and the campaign machinery behind them — see marketing operations.