Solving the challenge of understanding why certain creative assets perform well requires analyzing their visual elements. AI Image Decomposition uses Supermetrics Custom Fields to analyze the contents of your creatives, generating analyzable dimensions (like "sky" or "people") for your entire ad library.
Video tutorial: AI image decomposition use case
Key learnings and action steps
| Feature/Concept | What it does | Your action in Supermetrics Custom Fields |
| Auto-labeling | Automatically tags your images with general, out-of-the-box descriptors (e.g., "water," "cloud," "sky"). | Tag your library: Use a Custom Field to tag your entire ad creative library, allowing you to compare performance of assets with similar elements across platforms. |
| Custom prompts | Allows you to ask specific, strategic questions about the image composition. | Create binary dimensions: Ask questions like, "Does this ad contain people?" or "Is this a dark ad?" This returns a simple True/False dimension for filtering and sorting. |
| Strategic analysis | Converts previously unmeasurable creative elements into quantifiable data points. | Optimize creatives: Analyze the performance (e.g., CPA, CTR) of creatives based on these new dimensions to guide your creative team's future ideation. |
| Broad application | Applicable across product catalogs, creatives floating in DSPs, and media networks. | Standardize insights: Get better, data-driven answers on which visual elements drive high performance. |