Joshua Franklin Wiedeman

Creative Taxonomy

A creative taxonomy is a fixed set of tags applied to every ad: format, hook type, visual style, message angle, call to action, talent, and length. Without it, creative analysis is anecdotal (“the blue ad did well”). With it, you can compare groups (“UGC videos with question hooks had a lower CPA than studio videos with feature hooks”).

How it works

Most teams make many ads but have no consistent way to compare them. A taxonomy records the choices that differ between ads as tags that can be joined to performance data. Once every ad is tagged, you can cut results by any tag or combination of tags and see which ones perform best. These patterns are hard to see without consistent tagging.

How to do it

  1. List the variables that actually differ between your ads. Start with 5 to 7: format, hook type, visual style, message angle, CTA, talent, and length.
  2. Give each variable a short list of values that do not overlap and cover every case. For example, hook type: question, statistic, testimonial, before-after, pattern interrupt, trending audio.
  3. Put the tags in the ad name, for example VID_UGC_QUESTION_BENEFIT_SHOPNOW_CREATOR1_15S, so they come through with performance data.
  4. Build a report that groups results by each tag and lets you filter, for example “all UGC videos with question hooks.”
  5. Write creative briefs from the results: “Make a 15-second UGC video with a statistic hook and a benefit message.”

What to measure

  • Performance spread by tag: how much CPA, ROAS, or CTR varies across the values of each tag.
  • Winning combinations: how often a tag combination appears among your top ads.
  • Fatigue rate by tag: how fast performance drops for each creative type, used to set refresh timing.

Best practices

  • Keep the taxonomy small enough that the team tags consistently. A 20-field taxonomy nobody fills in is useless.
  • Automate tagging where you can: file names, platform labels, or a simple form that feeds a spreadsheet.
  • Review results monthly and update briefs. If the taxonomy does not change what you make, it is overhead.

Common pitfalls

  • Making values too fine to tag reliably, such as “slightly humorous” versus “moderately humorous.”
  • Comparing tags without accounting for spend and audience. A creative type can look better because it ran to warmer audiences.
  • Using the data in place of creative judgment. Data shows what worked; people still decide what to try next.