A pattern shows up in nearly every creative-heavy account we look at: hundreds of live creatives, filenames like final_v3_new and reels_copy_2, and no shared way to group them. Every performance question then becomes a manual hunt. Do our UGC hooks beat our founder hooks? Nobody can answer without opening the ad manager and eyeballing thumbnails.
The fix is not more dashboards. It is a creative tagging taxonomy: a small, shared labelling system applied before spend, so performance can be sliced by the attributes that actually explain it.
Why Creative Analytics Fails Without a Taxonomy
The core problem is that creatives are produced faster than anyone can organise them, and analysis needs structure that free-form filenames do not provide.
A creative tagging taxonomy is a standardised set of labels and naming rules applied to ad creatives, so that performance can be analysed consistently by attributes like hook, format, angle and offer.
The parallel is event data. Anyone who has built one knows a good event taxonomy is what makes in-app behaviour analysable, and that taxonomies are customer-specific: across Linkrunner projects, most teams track domain-specific event names rather than the generic GA4 defaults, because a copied scheme rarely fits. Creative tagging is the same idea applied one layer up, to the ads themselves.
When the tags exist, the payoff is concentration you could not otherwise see. Matiks, for example, found a single UGC creative concept drove around 60k installs, roughly a quarter of its total, a split that is invisible until creatives are grouped by type.
The Dimensions Worth Tagging
The instinct is to tag everything. Resist it. A taxonomy that is too detailed rots because no one maintains it. Tag the dimensions that change decisions:
- Concept vs variation. The single most useful split. A concept is the idea; variations are its cuts. You want performance rolled up to the concept, not scattered across 20 variations.
- Hook type. UGC, founder, testimonial, problem-solution, stat-led. Usually the strongest performance signal.
- Format. Static, video, carousel, playable. Different economics and different fatigue curves.
- Angle. The core message: price, speed, social proof, feature.
- Offer. The specific promotion or call to action, where one applies.
Optional secondary tags such as persona or duration are fine, but keep the required set to five or six. Every tag you add is a tag someone has to apply correctly on every asset.
A Naming Convention That Survives Scale
Tags need a carrier. A structured, delimiter-based naming convention encodes the key tags into the creative name itself, so the label travels with the asset wherever it goes.
- Use a fixed order and a delimiter. For example
concept_hook_format_angle_offer_version. Consistency matters more than the specific order you pick. - Keep values from a controlled list. ugc, founder, static, video. Free text is where taxonomies die.
- Decide manual versus automated early. Manual tagging breaks down exactly when volume rises, which is when you need it most. Element-level creative-analytics tools such as CreativeX, Foreplay, Motion and Segwise auto-detect creative attributes at scale, which is what makes tagging survive a heavy production calendar. This layer is creative intelligence: scoring the elements of the creative itself.
Whichever route you choose, the naming convention is the contract. If the name is right, the analysis downstream is cheap.
Connecting Tags to Performance and Revenue
Tidy names are not the goal. Tags are only useful when they join to spend, installs and revenue, so you can compare like with like.
- Push the tag metadata into your reporting layer. The creative's tags need to sit alongside its spend and outcomes, not in a separate spreadsheet.
- Join through to revenue, not just installs. A hook that wins on CPI can lose on ROAS. The tag is only decision-grade once it resolves to payback. This is where an attribution layer such as Linkrunner earns its place, joining tagged creative metadata to installs and revenue.
- Define the minimum cut. Tag dimension against ROAS, CPI and retention is enough to start reallocating. The guide to building a creative performance dashboard covers the layout.
Do this and "UGC beats founder on ROAS by a clear margin" becomes a query, not an argument.
Rolling It Out Without a Data Team
A taxonomy only works if the creative team can follow it without a data engineer in the loop. Keep the rollout light:
- Write a one-page spec. The tag list, the allowed values, the naming order. One page, not a wiki.
- QA before spend, not after. Catch untagged or mistagged creatives at upload. A creative that goes live mistagged pollutes the analysis for its whole flight. The discipline mirrors the QA stage in the event taxonomy implementation guide.
- Review on a cadence. Revisit the tag set each quarter. Retire dimensions nobody queries and add ones the team keeps asking for. A taxonomy that never changes is one nobody trusts.
For the creative principles that make the tags worth analysing in the first place, the ad creative optimisation guide is a useful companion.
FAQ
What should a creative tagging taxonomy include?
At minimum: concept versus variation, hook type, format, angle and offer. Keep the required set to five or six dimensions so it stays maintainable, and draw values from a controlled list rather than free text.
Manual tagging or automated creative tagging, which is better?
Manual works at low volume and breaks exactly when production scales up. Automated, element-level tagging holds up under heavy output, which is when a taxonomy matters most. Many teams start manual and move to a creative-analytics tool as volume grows.
How many creative tags is too many?
If people stop applying them correctly, you have too many. Every dimension is a field someone must fill on every asset, so more tags mean more decay. Five or six well-chosen dimensions beat fifteen that go half-filled.
How do creative tags connect to ROAS and retention?
Tags become decisions only when they join to spend and revenue in your reporting layer. Once tag metadata sits alongside installs and revenue, you can rank hooks, formats and angles by payback rather than by clicks.
Closing
A creative tagging taxonomy is unglamorous infrastructure, and it is the difference between an account you can analyse and one you can only guess at. Keep the tag set small, encode it in a naming convention the creative team can actually follow, QA before spend, and join the tags to revenue so they drive reallocation rather than tidy folders.
If you want the tags to resolve to payback automatically, Linkrunner joins tagged creative metadata to installs and revenue in one reporting layer. Request a demo, or start smaller: write the one-page spec this week and tag every new creative before it goes live.