
Use a governed structured field when a media library must accept one standardized campaign category. Free-text tags do not enforce that vocabulary. In Cloudinary, structured metadata fields are defined globally with types and optional validation, while tags and contextual values remain unvalidated text.
Key Takeaways
- A useful label is not automatically a validated field.
- Field definitions and per-asset values have different owners and scope.
- A valid category still does not verify the pictured product.
Choose the record from the required decision
Cloudinary's custom-metadata reference distinguishes tags, contextual metadata and structured metadata. Structured fields can use strings, numbers, single- or multi-select lists and dates, with optional validation. Their definitions apply across the product environment; values are set for individual assets.
Tags are useful for flexible grouping. They are not a substitute for a controlled category when a publishing integration expects one exact value.
Design one small classification contract
- Identify the actual decision, such as campaign, product family or approval state.
- Record the approved vocabulary and who can change it.
- Have the Assets administrator or responsible developer define the appropriate structured field through the documented management route.
- Assign values to a non-sensitive sample set using authorized asset permissions.
- Check that the receiving selection or publishing workflow reads the intended field.
- Inspect the chosen assets visually before approving the classification.
Cloudinary says global fields are managed by an Assets administrator or through its metadata API. Asset write permission lets a user edit values; it should not be confused with authority to redesign the organization-wide schema.
A hypothetical catalogue inconsistency
A team uses summer-sale, summer_sale and Summer Sale as informal tags for the same campaign. A downstream rule expecting one exact value may not produce the intended set.
The proposed repair is a documented field and vocabulary, followed by a reviewed mapping of existing records. Do not automatically merge every similar-looking tag: two campaigns may intentionally have similar names.
The product-to-SKU guide supplies an independent content check. A correctly formatted campaign or SKU value does not prove that the image shows the correct variant.
Verify schema and content separately
Record the field definition, allowed values, sample assignments and actual receiving behavior. If a value fails validation, identify the field rule rather than removing validation across the environment to accept one entry.
Official ImageKit upload-settings guide likewise requires custom fields to be created before assigning upload metadata. That is a cross-provider setup principle, not a claim that its schema controls or request syntax match Cloudinary's.
Accept when the library expresses the agreed category consistently and the intended assets appear in the receiving workflow. Preserve a record of any migration exceptions. This is a documentation-based classification plan, not an executed schema change or a claim that database fields are embedded automatically in downloaded photographs.
Related tasks
- How Do I Keep AI Product Images Matched to the Correct SKU?
- How Do I Keep the Approved AI Image or Video from Getting Lost Among Revisions?
- Lightroom Metadata Conflict: Keep the Disk or Catalog Version?
Sources and scope
- Cloudinary: Custom metadata: Tags and contextual values are unvalidated text; structured fields have global definitions, types and optional validation; field-definition and asset-value permissions differ.
- ImageKit: Upload files and folders: Overwrite settings independently control AI tags, user tags and custom metadata; fields must be created before assigning custom metadata on upload.
Documentation-based editorial synthesis with AI-assisted research and drafting. Scenarios, ledgers and acceptance checks are hypothetical or proposed. No hands-on test, account change, measured demand, AI-citation rate, benchmark or client acceptance is claimed. Availability and plan limits remain within each cited source's scope.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.