For performance creative teams running variant tests

One master concept, 20 to 40 platform-ready variants

The core production workflow for the Performance Creative Rotation Playbook. From approved master to platform-spec compliant variants in about 90 minutes.

Open workflow

When to use this recipe

Built for performance creative teams running variant tests at meaningful spend. Use it when these four conditions hold.

90 mFrom approved master to platform-spec compliant variants in about inutes
20 to 40A repeatable workflow that turns one approved master ad concept into...
20 to 30 mFilter to the ost-promising combinations

Master concept is approved and locked

Variants from a moving master waste production. Lock the master first; then run variants. Otherwise revisions cascade across the matrix.

You need 15 or more variants

Below 8 variants, the workflow overhead does not pay back. Above 15, the batch workflow comes alive. Sweet spot is 20 to 30 per campaign.

Variants must stay strictly on-brand

Brand kit, style references, and character LoRAs lock the brand layer so junior team members can run variants without drift. Senior creative does QA, not execution.

Variation along specific test dimensions

Headline copy, framing, CTA, casting, aspect ratio. The matrix is testable; the variants are repeatable. This is variant testing, not exploratory creative.

The workflow

Seven steps from approved master to platform-ready variant set, with realistic time estimates.

1
Build the variant matrix (15 minutes)
Open a spreadsheet. Define dimensions (aspect ratios, headline options, CTA options, casting variations). Multiply to size the matrix. Filter to the 20 to 30 most-promising combinations.
2
Open the variant generator workflow (5 minutes)
Navigate to your saved workflow templates and open Campaign Variant Generator. The workflow has brand kit, style references, character LoRA, and master concept input slots pre-locked.
3
Load and validate the master (5 minutes)
Drop the master concept into the workflow. Verify brand-lock variables read correctly. Run a single test generation with defaults before scaling to the matrix.
4
Run the matrix (45 to 60 minutes)
For each row in your variant spreadsheet, configure inputs and run the generation. Run 5, review, refine prompts if needed, then run the rest. Image variants average 8 to 15 seconds each.
5
First-pass review (15 minutes)
Brand consistency check. Spec compliance check (aspect ratio, text). Obvious quality failures (hand errors, text errors, character drift). Reject and regenerate failures.
6
Senior creative QA (10 to 15 minutes)
Pull the 20 to 30 surviving variants into a review pass. Senior scans for variants that are technically correct but creatively off. Approval rate typically 60 to 80%.
7
Format and label for ad platform upload (10 minutes)
Export per platform spec. Name files to your media buyer's tagging schema. Hand off to paid media with the variant matrix attached for performance tracking.

Tips and failure modes

Six patterns that separate teams who run variants smoothly from teams who hit the same issues every campaign.

Lock brand kit BEFORE running variants

Brand colors, type, logo placement, character identity. The locks are why juniors can run variants without senior review per asset.

Filter the matrix before generating

A 4x5x3 matrix is 60 theoretical variants. You do not need all 60. Pick the 20 to 30 that test actual hypotheses.

Run 5 then review before scaling

If you batch 30 and the first 5 share a subtle quality issue, you have wasted 25. Run small, review, then scale.

Verify text in image character by character

AI text rendering is improving but not perfect. Brand-mark errors are reputationally damaging. Read every text element before approving.

Name files for media-buyer tagging

Variant_AspectRatio_Headline_CTA naming makes performance attribution work. Without it, your test data is a mess and the campaign learnings die with the campaign.

Senior QA is non-negotiable

Junior can run the workflow; senior must approve the cut. Brand drift catches most often in the senior pass, not the technical pass.

Frequently asked questions

What performance creative teams ask before adopting the variant generator workflow.

About 90 minutes for 20 to 30 variants once the workflow template is set up. Matrix planning 15 min, generation 45 to 60 min, review and labeling 15 to 20 min.

Yes, once a senior has built and locked the workflow template. The brand kit and style locks are the safety rail. Senior reviews the final cut; junior runs production.

Same workflow shape, longer compute. Video variants average 30 to 90 seconds each. 30 variants takes 15 to 45 minutes of compute time plus configuration.

Above 40 to 50 per campaign you usually hit diminishing returns on testing. Pick the 20 to 30 hypotheses worth testing rather than maximizing variant count.

Stop. Re-lock the master. Rerun the variants from scratch. Trying to patch existing variants to a new master creates inconsistency that breaks the test.

Yes, with smaller matrices. Organic variant counts are usually 5 to 10 per post instead of 30. The workflow shape is the same; the matrix is smaller.

Name files for your media-buyer tagging scheme so performance attribution links back to the variant matrix. Otherwise you ship the test, get the data, and cannot tell which dimension drove the lift.

Yes, when brand kit and style locks are properly configured. Enterprise teams also use workflow-level permissions so only senior creatives can edit the locks while juniors run variants.

Open the workflow and ship your variant batch today

Campaign Variant Generator is a saved workflow template with brand kit, style references, and character LoRA slots pre-configured. Load your approved master and run the matrix.

Open workflow
DESIGNERBOX
×