Creative Strategy

One Source Asset Can Become Dozens of Variations. Approve the System, Not Just the File.

One approved source asset branching into four visual derivatives beside a navy creative strategy editorial panel.

Updated: October 2026

Demand Gen asset automation can generate design variations and videos from existing creative, so approving the source file is no longer enough. Approve the inputs, permitted transformations, claims, placement constraints and rejection rules. Then sample the derivatives that can actually serve and measure them as a system.

What Google Ads API v26 exposes

Google’s v26 release notes describe Demand Gen support for asset automation that generates design variations and videos from existing assets. The same release expands reporting and control surfaces across automated creative. This makes derivative governance a practical creative-operations requirement, not an abstract AI policy.

The source asset remains important, but the customer may see a crop, layout, motion treatment, sequence or text-image pairing that no one exported from the original design tool.

Create a source-of-truth package

Input Lock Allow
Product Shape, color, packaging, quantity Background and framing
Message Supported claims and qualifiers Order and concise phrasing
Brand Logo, colors and prohibited contexts Placement-native composition
People Identity, rights and sensitive traits Approved crops
Motion No false product behavior Pacing and transitions

Approve transformation rules

Write down which operations automation may perform: resize, crop, background extension, animation, video assembly, text reflow, music or voice. For each, define the defect that stops delivery. A missing disclaimer, invented button, altered label, unsafe crop or impossible demonstration should not wait for performance data.

Design for derivative resilience

Keep key subjects away from edges, separate text from busy imagery, provide clean product references and use captions that survive vertical and square formats. Build creative so adaptation has room to work without destroying hierarchy.

Use a sampling plan, not a perfect-coverage fantasy

Automation may produce more combinations than a team can review exhaustively. Prioritize the highest-spend placements, the most material transformations and the assets with people, products, prices or regulated claims. Review a representative sample before launch and capture served examples after launch.

Pair visual review with MTC’s cross-format QA workflow and AI video continuity checklist.

Measure the system without losing the idea

Separate the creative hypothesis from the adaptation layer. If one concept is supplied in many automated formats, label the shared message angle and the generated treatment. Compare qualified outcomes at concept, source asset, format and placement levels where reporting allows.

  • Concept outcome: did the promise resonate?
  • Derivative outcome: did adaptation improve delivery or clarity?
  • Placement outcome: did the asset fit the interface?
  • Quality outcome: did support, complaints or returns change?

Do not declare that “AI creative won” when the real difference was simply a better aspect ratio or more inventory.

Build an incident-ready rollback

Record the source asset IDs, automation settings, approval date and reviewer. Define how to disable the automation or remove a defective derivative without pausing every campaign. During the first days of delivery, review placement previews, customer comments, product-page engagement and disapprovals.

Maintain a derivative register

For each source asset, keep a compact register of the active campaign, automation types, eligible placements, first-seen date and reviewed examples. The register does not need to store every possible frame; it should let the team trace a served variation back to its source and settings. Add a defect log with severity, disposition and recurrence so repeated failures influence future input design.

Review the register whenever the source changes. Replacing a product image, offer or legal line can invalidate prior derivative approval even if the campaign name is unchanged. Archive old source packages so the team can explain which creative existed for a past complaint, test or reporting period.

Frequently asked questions

Should asset automation always be disabled for brand-sensitive campaigns?

No. Use it when the permitted transformations and monitoring match the risk. Disable specific operations that cannot preserve required fidelity or disclosures.

Can performance data replace creative QA?

No. A misleading or defective asset may still attract clicks. Quality and substantiation are launch requirements, not optimization metrics.

Who owns automated derivatives?

Creative owns brand and fidelity, media owns settings and delivery, and the business owner owns product and offer truth. Assign named approval rather than shared ambiguity.

Sources

Published by Marketing That Clicks
Last reviewed October 2026.