Braze naming conventions that survive a Friday afternoon

Picture the lifecycle manager — the person responsible for the welcome emails, retention nudges, and win-back sends a brand fires off automatically — at the end of a long week. One campaign left to set up before they log off. Naming doc says twelve dimensions, dropdown-heavy, ninety characters out the other side. Structured way takes forty seconds and three context switches to a Notion tab. Free-text takes four. Guess which one ships.

That is the usual cause of death. Not bad rules — slow rules. The convention was thorough, defensible, exhaustively documented. Nobody used it, because using it cost more than ignoring it. Six months on, nobody can locate the asset that fixed the churn bleed last Q3. (Asset is the catch-all term Braze uses for any nameable object — a campaign, a journey, a saved audience, an email template, a reusable content block.)

Conventions that survive share three traits: they fit on a sticky note, compress to a readable 40–60 characters, and their dimensions map one-to-one to the filters you actually use in reports. If a dimension exists in the convention but does not exist as a filter in reports, drop it. You are carrying weight you will never use.

Second failure mode: enforcement by documentation. A beautifully written Notion page never moved compliance. Slack reminders do not either. The only thing that works in practice is making the structured way faster than the free-text way — a naming tool open next to the Braze tab, keyboard-accessible, one-click-to-copy.

Before the rules: the filters. Almost every report you will ever build in Braze starts the same way — pick what kind of thing you are looking at, then which channel, then which program, then which audience. Four cuts. Match those four in the name and report-side filtering comes for free. Anything past the four is decoration unless it earns its place.

Asset Type — campaign, canvas, segment, template, content block. Anchors the name to what the object is inside Braze. Quick gloss because these terms are not universal: a campaign is a single send (one email, one push); a canvas is Braze's name for a multi-step automated journey (welcome series, dunning sequence — dunning being the polite word for chasing failed payments); a segment is a saved audience definition; a content block is a reusable snippet of HTML or copy. Without the asset type prefix, campaigns and canvases blur together in any search list longer than a screen.

Channel — email, push, SMS, in-app, content card, webhook. Most lifecycle reporting splits by channel first, and it is the filter most team members reach for before anything else.

Program — onboarding, activation, retention, dunning, win-back, feature adoption, upsell, re-engagement, transactional, promotional. These are lifecycle stages — phases of the customer relationship the message is trying to move — not campaign names. Program is the family; the specific campaign is a suffix.

Audience — all, free, paid, trial, churned, at-risk, new, dormant, VIP. Segment targeting shows up in every reporting query, and pulling it into the name means you never need an ad-hoc segment join to slice results.

The canonical form that works: [asset]_[channel]_[program]_[audience]. Example: campaign_email_onboarding_newuser. Forty characters, every component matches a real report filter, no decoration. Audience names come from your segmentation model — the segmentation guide covers how to derive them so the vocabulary is shared across every program.

One small formatting rule worth committing to. Underscores between dimensions, hyphens within a dimension, so campaign_email_onboarding_welcome-series_newuser is parseable both by eye and by script. Which separator matters less than consistency — pick one and never mix.

The Orbit Namer — a small web tool that produces canonical Braze names from dropdowns — exposes six dimensions past the core four: Country, Language, Version, Step, Variant, and Deployment Date. These are not always wrong. The failure mode is using all ten always rather than using them selectively.

Country and Language earn their place when the program runs in parallel per market with separate sends. Drop them when it is one global send. Version earns its place when a v2 redesign is running alongside v1 and both need to be measurable. Step earns its place on sequenced journeys (day-1, day-3, day-7) where each send is independently tracked in reporting.

Variant belongs in the name for structured A/B tests — splitting your audience to compare two or more versions — where the variant letter (A/B/C) is the analysis key. Deployment Date is the trickier one. For recurring programs, dates belong in tags, not names — putting them in the name breaks historical reporting every time the program refreshes. For one-off promotional sends (black-friday-2025, summer-sale-2026), the date is part of the asset's identity and belongs in the name itself.

The split most teams get tangled on. The name answers: what IS this asset? Tags — Braze's freeform labels you can attach to any asset for filtering inside the Braze UI — answer: how do you want to group it? Identity in the name. Grouping in tags. That is where most naming-convention failure happens in practice.

Date-sensitive promotions (black-friday-2025), experiment variants (test-hero-v2), team ownership (growth-team, lifecycle-team), and campaign cohorts (q1-launch-batch) all belong in tags rather than names. Tags filter inside the Braze UI; names are what reports match on. A tag is metadata a campaign CAN have. The name is what the campaign IS.

A concrete test that always works. If you renamed the asset, would the report still make sense? Yes — filterable value is stable metadata, so it is a name. No — it changes across instances of the program, so it is a tag. The Orbit Namer skill surfaces recommended tags for every asset it names, so the split is done for you rather than relitigated in Slack every week.

Documentation-based enforcement fails. Slack reminders fail. Quarterly naming-compliance reviews fail too — offenders are long gone by the time the audit lands. The one approach that works in practice is making the structured path faster than the free-text one.

Concretely: a naming tool open next to the Braze tab, pre-populated with the dropdowns your team filled in on day one, producing the canonical name plus recommended tags in a single click. The Orbit Namer is built for exactly this — six seconds from intent to copied name.

For larger programs, add a gate: a pre-launch QA checklist that verifies the asset name matches the convention regex — a short pattern-matching expression — before the campaign is approved. Braze does not ship this natively. The name field is free-text, no built-in validation, no programmatic enforcement. The Braze Instance Audit skillscans the workspace for off-convention assets and produces a prioritised cleanup list on demand. If you want to rename the worst offenders in bulk, Braze's REST API — the HTTP interface for programmatic changes — supports it, and renames are retroactive: historical data stays linked to the new name automatically.

One adoption rule for messy existing workspaces: do not rename everything at once. Enforce on new assets first. Audit once a quarter. Bulk-rename the top 50 most-reporting-critical assets only after you have a full quarter of clean new-asset data — that is when the baseline for good is clear enough to project backwards. Rush it and you will rename things twice.

The one thing to do Monday: open Orbit Namer in a tab next to Braze and use it for every new asset for a fortnight. Do not announce a policy. Do not write a doc. Make the structured path the path of least resistance and watch what your team does without being asked.