Prompt engineering is an essential skill for marketers, guiding the content delivered by generative AI tools such as ChatGPT and DALL-E.
Prompt engineering might sound technical but refers to the crafting of inputs that instruct the AI tools. Similar to providing a brief to a client or agency, a prompt is the direction you use to communicate what you want an AI system to produce. Like a request, a prompt needs to be engineered to produce the response you need.
A great prompt clearly defines the context, goal, format, and tone, ensuring that the AI tool understands what you’re aiming to achieve. By learning how to iterate on prompts and fine-tune them for tone and audience, marketers can harness AI to produce consistent, high-quality content that aligns with brand voice and goals.
A request for a “campaign for an energy drink” will produce different output, compared to a prompt for a “fun, smart, ironic ad copy for a new energy drink for 16–25-year-old males with the supporting imagery attached for reference”.
Iterating is the key to prompt engineering. It is unlikely that the system will produce a perfect response on the first attempt. As the prompt engineer, you need to have a clear vision of what you need to be able to massage the prompt to deliver your desired outcome.
Table 1 demonstrates how to effectively prompt an AI tool to achieve your desired outcome.
Table 1: Prompt examples
Key takeaways:
When prompting, you should ensure that you:
- Be specific
- Revise and iterate your prompt if you don’t get the desired result
- Ask yourself if the prompt output and outcome follow your company’s chosen governance framework
Exclusively available to ADMA Members, the ADMA AI Toolkit is a marketer-ready guide on how to use AI effectively and responsibly – developed by industry for industry
