A Generative Approach to Brand Strategy
We fine-tuned an AI model to generate "brand pyramids" instantly.
As someone who’s spent a long career in advertising, I’ve always been mystified by the brand strategy process. There is an art to it. And there’s a science. But finding alignment is often fraught with challenges. Things like time commitments, flawed research methods, human bias, and conflicting stakeholder feedback can make brand strategy projects messy and ineffective.
At Addition we’ve been exploring new methods for brand strategy that leverage the generative capabilities of foundational AI models like GPT-3.
Here’s how it works:
We tap into the massive amounts of web content that GPT-3 has been trained on, including its innate understanding of the relationship between brands, people, and culture.
We define a framework for the output (in this example we’ll use a “brand pyramid” template).
We fine-tune the model by providing it with high-quality input and output parings using that framework.
Once this is complete, our AI model can generate brand strategies with a single prompt: “Recommend a brand pyramid for ________”
While this is not a replacement for the nuance or creativity of a brand owner or strategist, these techniques can play a supporting role by helping marketers and their partners workshop and align around strategic directions faster.
Let’s see it in action. The following content was generated by a 🤖 with no-human editing.
Brand Pyramid for Heinz Ketchup
Brand Pyramid for Tesla
Brand Pyramid for Levi’s
Brand Pyramid for Minecraft
Brand Pyramid for Hamsters
Brand Pyramid for Chuck Norris
Want to learn more?