Every marketing organization eventually has the meeting. A new naming convention is presented: clean, logical, built with care. The analytics team likes it because campaign performance might finally roll up without a week of manual re-mapping. IT signs off because it respects the character limits and metadata fields their platforms actually support. Then someone from the creative or brand team raises a hand and says the quiet part out loud: “Most of what my team makes doesn’t fit this template.”
In one recent engagement, a client’s creative lead estimated that roughly 90% of her team’s day-to-day output fell outside the convention that had just been presented as finished. That work included retail signage in non-standard sizes, internal collateral, and channel-specific one-offs. Her own caveat, “that may be extreme,” didn’t change the point. She wasn’t being difficult. Instead, her feedback reflected how the team works, and it made the taxonomy stronger.
This is the reality of taxonomy governance. Naming conventions sit at the intersection of marketing, data, and IT, and each function has requirements that often conflict. The question isn’t whether the three will disagree, because they will. The question is who decides when they do, and most organizations have never answered it.
Why Naming Conventions Become Battlegrounds
A naming convention looks like a small thing, a string of underscores and abbreviations. In practice, it is an operating agreement between three groups optimizing for different outcomes:
- Marketing (including creative and brand teams) optimizes for speed and readability. Campaign teams work against launch calendars. They want names a human can scan in a media platform at 11 p.m. before a flight date, and they want to call a moment “HolidayBlitz” without filing a ticket.
- Data optimizes for consistency and machine-readability. Analytics teams need every field to parse predictably because the taxonomy is what lets spend, delivery, and outcome data join across platforms. One rogue delimiter or free-text field turns an automated pipeline into a weekly manual cleanup.
- IT optimizes for system constraints and durability. Platform owners know the character limits, the reserved characters, and which fields sync between the digital asset management system (DAM), the customer data platform (CDP), and the ad servers and which don’t. They’ve also seen conventions change three times in four years, and they carry the migration debt each time.
None of these positions is wrong. The trouble starts when ownership is ambiguous, because the convention then defaults to whoever built the spreadsheet, and the team slides into a “whoever cares most” model of governance. As a result, the first time a launch deadline collides with a validation rule, the convention starts to erode.
The Wrong Answer: A Single Owner
The instinctive fix is to appoint one owner, and each choice fails in a predictable way. Give it to Data, and the convention becomes technically perfect and practically ignored, because creative and campaign teams experience it as bureaucracy. With Marketing in charge, it can drift toward human-friendly names that break downstream automation within a quarter. Put IT in charge, and it can calcify, becoming rigid, slow to change, and disconnected from how campaigns naturally get planned.
Single ownership fails because a taxonomy is not an asset that belongs to one function. Instead, it is shared infrastructure, closer to a contract than a document, and a contract is governed by the parties who depend on it rather than owned by one of them.
The Better Answer: Taxonomy Governance Through Decision Rights, Not Ownership
Organizations that make taxonomy stick reframe the question from “who owns it” to “who decides what.” In practice, that looks like a lightweight federated model with four parts:
- A cross-functional taxonomy council holds the structure. A small standing group with marketing operations, analytics, and platform/IT representation, and creative at the table, owns the architecture of the convention: the tiers, field order, delimiters, and governed picklists. It meets on a set cadence, not ad hoc.
- Each function holds decision rights over its own domain. Marketing decides campaign and initiative names within the agreed structure. Data defines the fields that must remain machine-readable and the values in governed picklists. IT holds a veto on anything that would break a platform constraint. Nobody redesigns the convention unilaterally.
- A single, versioned data dictionary is the source of truth. Not five spreadsheets in five drives, but a single dictionary with one version history, where a new hire or agency partner can find out what “CTV_15s_ProsQ4” means.
- An explicit escalation path handles real conflicts. When speed and structure collide, the tie-breaker should be agreed in advance, typically a marketing operations or data governance lead who is accountable to both sides. That way disputes resolve in days rather than release cycles.
Five Lessons From Taxonomy Work in the Field
1. Scope tightly and govern the campaign spine first. The fastest way to kill a taxonomy is to force every asset in the enterprise into it on day one. Start with the activation assets that flow into paid, owned, and earned channels: the campaign, placement, and creative naming where measurement value is highest. Defer the long tail (internal collateral, one-off formats, non-campaign work) to a documented second phase. Stating the boundary openly is what earns the creative team’s trust.
2. Encode the rules where the work happens. A PDF style guide is a suggestion, while a picklist is a policy. Wherever possible, enforce the convention at the point of entry: dropdown values in the workflow tool, validation in the ad platform, required metadata fields in the DAM. If compliance depends on people remembering a document, you have hope rather than governance.
3. Keep the leading fields consistent. Even teams whose work doesn’t fit the full template can usually adopt the first fields (brand, campaign, date) consistently. That partial consistency still makes most assets findable and joinable while the edge cases are worked out. Don’t let an unfinished tail hold up the fields that already work.
4. Design for the least technical user. Coordinators, designers, and agency partners will apply the convention at speed, not the data engineers who designed it. If it can’t be followed correctly under deadline pressure, it will be followed incorrectly. Fewer fields, governed values, and sensible defaults beat exhaustive precision every time.
5. Treat changes like product releases. Taxonomies evolve with new channels, formats, and business units. Version the dictionary, announce changes with lead time, and never change historical values retroactively without a mapping plan. Every unmanaged change is a future reporting break.
Why This Matters Beyond Reporting
Inconsistent naming has always taxed organizations quietly through hours of manual re-mapping, campaigns that can’t be compared, and assets no one can find. Now, that cost is rising. AI-enabled content operations, automated campaign orchestration, and agentic workflows all depend on metadata that machines can trust, and an AI agent can’t reliably assemble, tag, or measure creative it can’t parse. Taxonomy governance is now part of the infrastructure AI depends on, and the organizations investing early in clear decision rights, governed dictionaries, and phased rollouts are the ones whose automation will work.
Nobody should own your naming conventions outright. Not Marketing, not Data, not IT. A cross-functional council should hold the structure, the teams closest to each domain should govern it, and conflicts should resolve through decision rights agreed in advance rather than whoever escalates loudest. Scope the first phase to where measurement value is highest, enforce the rules in the tools rather than in documents, and treat every change like a release.
Do that, and the meeting where someone says “most of our work doesn’t fit” becomes useful input to a system built to evolve, rather than a crisis.
Transparent Partners helps enterprise marketing organizations design and operationalize build the data, technology, and governance foundations that make AI-enabled marketing work. If your taxonomy has become a turf war, we’d love to talk.
Anoop Reddy Yeddula, Analyst
