Exemplar-to-Creative Workflow
Reverse-engineer winning examples before generating original campaign work
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 99%
This workflow begins before creative generation. Collect several high-quality advertisements or campaigns and ask an AI analyst to explain why each one succeeds, including its simplicity, emotional tension, visual symbolism, brand association, offer clarity, and call to action. Compare the analyses to identify common mechanisms rather than superficial stylistic similarities. Turn those mechanisms into requirements for a new creative brief, then pass the brief to an image or copy generator. Finally, evaluate the generated work against the same criteria extracted from the exemplars. The workflow improves output because the model receives a concrete definition of quality and a structured set of constraints instead of a generic request to make something compelling. It also catches basic omissions that enthusiasm for a concept can obscure.
Origin
Extracted from Marketing Against The Grain through Kipp Bodnar's analysis of Netflix's Lupin campaign and its connection to AI creative generation.
Core principles
- 01Study excellent work before asking AI to create
- 02Extract mechanisms rather than copying surface appearance
- 03Find common elements across multiple exemplars
- 04Feed the findings into the generation brief
- 05Preserve marketing basics alongside creative novelty
How to run it
- 1
Select exemplars
Gather three to five pieces of work that demonstrably communicate well to a comparable audience. Favor examples with different executions but similar objectives.
Pro tip Include examples you can explain as successful, not merely ones you personally like.
Watch out Poorly matched references can teach the wrong lessons.
- 2
Deconstruct each example
Ask AI to identify the emotional, visual, strategic, and conversion elements that make every example effective.
Pro tip Request observations about both creative subtlety and basic information such as timing and destination.
Watch out Do not accept vague praise such as “engaging” without a mechanism.
- 3
Extract common mechanisms
Compare the analyses and identify the elements that recur across the strongest work. Translate them into brief requirements.
Pro tip Separate transferable principles from brand-specific execution.
Watch out Copying imagery or language can create derivative work and legal risk.
- 4
Generate and score new work
Use the extracted requirements to generate original creative, then assess whether each requirement appears in the result.
Pro tip Ask one AI process to analyze and another to generate for a cleaner handoff.
Watch out A visually attractive output can still omit the offer or call to action.
In the wild
Kipp supplied ChatGPT with a minimalist Lupin advertisement showing a wrist where a watch had been stolen. The model identified simplicity, mystery, emotion, luxury cues, brand association, and the call to action. He proposed using those findings as the pre-step for briefing a new campaign.
→ A successful advertisement became a reusable set of creative requirements rather than a one-off inspiration image.
Common mistakes
Copying surface style
Replicating colors or composition without understanding the underlying communication mechanism produces derivative, fragile work.
Using only one example
A single exemplar makes it difficult to distinguish transferable principles from accidental features.
Forgetting marketing basics
Creative intrigue cannot compensate for a missing offer, release date, destination, or call to action.
Is it for you?
Best for
Marketers developing ads, messaging, or campaigns in a category where strong reference examples are available.
Not ideal for
Situations where the selected examples are irrelevant to the audience or are being copied rather than analyzed.
From the transcript
“So go and just gather three to five great ads and then ask ChatGPT to break down for you.”
“Take some of the best work, find the common elements, and then make sure that those common elements are included in the new work and…”
“This would be like the pre-step to what you just showed, right?”
From the episode
6 AI Growth Hacks Top Marketers Don't Want You To Know (#167)