Using AI To Write Better Briefs

Contagious posted on how to sharpen your briefs using Gen AI with The Liberty Guild, Congregation and the brilliant Zach Duenow.

(A 3-minute read with an oat flat white)

First published by Contagious


Creative briefs are effectively prompts and generative AI can make them better, according to The Liberty Guild.

The decentralised creative agency has developed a process for using generative AI to make briefing more effective and efficient, and in a webinar for the World Federation of Advertisers in February, it gave an abridged explanation of how the process works.

Tim Sparke, the co-founder of Congregation Partners who worked with The Liberty Guild to develop the process, began by citing a 2021 survey, which found that while 78% of marketers think they write good briefs, only 5% of agency staff agree.

He also noted that while AI has been used at the beginning of the marketing process, to analyse sales data and customer behaviour, and then to produce assets at the end, it hadn't been applied to the in-between stages, to develop and refine strategies and briefs.

But Sparke believes that using generative AI can help marketers think more laterally and improve the quality of their writing, as well as make the briefing process more efficient.

To begin, you have to prime the AI. Large Language Models like ChatGPT know too much, and strategy is about ‘less’, said Muze co-founder Zach Duenow, who was also presenting on behalf of The Liberty Guild.

So, feed the AI documents about your brand and business challenge, and then test it with a series of questions about your positioning and customer profile to make sure that it has understood everything.

Next, you’ll want to prime yourself. The Liberty Guild recommends asking the LLM, for example, to ‘act as Malcolm Gladwell’ and explain five big trends in your product category, and to list 10 marketing clichés from the sector, to get in the right mindset.

As for the brief itself, The Liberty Guild has developed an iterative, three-stage process for using AI to turn commercial problems into marketing opportunities that creative agencies can really get their teeth into.

Start with a basic but extensive prompt that reminds the LLM to use the information from your brand documents, and sets out guardrails to mitigate against common marketing errors, to write a brief that reframes the problem statement in a way that inspires creative teams.

From there, use ‘few shot prompting’ (e.g. telling the AI to try again, focusing on the bits you liked and ignoring the bits you didn’t) to improve its initial responses.

When you think you have the kernel of a good brief, you can polish it by using advance prompts, which essentially ask the AI to roleplay in order to find original angles. For example, you could tell the AI to imagine that it has hypnotised a consumer, who then reveals their subconscious thoughts and reservations about a product or category. Or you could tell it to imagine that it’s a behavioural economist and come up with 10 nudges that might get people to consider your brand.

This process — combined with social listening, using AI to digest and synthesise product reviews and forum discussions — describes the basic rhythm that can be applied to pretty much every aspect of the brief, from defining the product benefits to describing customer behaviours.

The key to the process, according to The Liberty Guild, is the dialogue in the prompting. Use imaginative scenarios to elicit novel responses from the LLM, and keep it on track by constantly telling it what's good and what isn't. It's the back and forth combining human judgement and experience with AI's almost limitless capacity for generating suggestions that produces valuable results. AI can’t replace people in the briefing process*, and it can’t write for you, but it can be a useful tool for creating better briefs.

* yet, anyway.

First published by Contagious

Photo by Christin Hume on Unsplash

Previous
Previous

Podcast: How To Be Kind At Work, Even When You’re Under Pressure

Next
Next

Are Leaders Underestimating The Impact Of Generative AI?