A deep dive into writing prompts for generative AI to create an email campaign brief.
In first part of a two-part article, Kath Pay takes a deep dive into writing prompts for a genAI-created campaign brief. In the second part, we’ll see how the LLM responded.
Persuasion is at the heart of so much of what we do as marketers, no matter which channel we’re working in. As marketers, we can be most effective when we understand why we humans do what we do and use that information to help our customers make decisions others make on matters that concern us.
In marketing messages, everything you do can benefit from effective persuasion – knowing how to frame an argument or request in the best way to draw the response you want, from clicking through to a landing page or converting.
Persuasion principles and cognitive biases
In my last MarTech post, I showed you how to combine email design with several of Robert Cialdini’s seven persuasion principles, such as loss aversion and social proof, to make your emails more click- and conversion friendly.
In this post, I’ll review how to write message copy that combines Cialdini’s persuasion principles and the cognitive biases that are part of human nature and which we use as shortcuts to help us make decisions.
This is not a license to try to manipulate your customers into doing what you want them to do. Rather, your goal should be to help customers make decisions that will allow them to solve a problem, satisfy a need, or achieve a goal.
You’re speaking to people who, by virtue of opting in to your email program, are seeking you out. If you can help them achieve their needs, they in turn will help you.
You also can use GenAI in your A/B testing. I’ll explain how to do that in my next MarTech post.
Where generative AI fits in
I’m framing this article by using generative AI to create your first round of creative briefs and message drafts. I’ve used ChatGPT, but you can substitute Google’s Bard or any other large-language model tools that you might be familiar with already.
However, be sure you are well-grounded in the principles behind persuasion and cognitive bias before you start using your AI tool. This is a must. Don’t expect ChatGPT to understand what you’re asking it to do. Check these resources:
- If you don’t already have a good background in psychology in marketing and behavioral science, start by reviewing Cialidini’s own thumbnail guide to his seven principles of persuasion, or read his landmark book, Influence: The Psychology of Persuasion.
- See some examples applied to email marketing in my MarTech post, Why and how to use loss aversion in email marketing.
- Get a quick introduction to cognitive biases from Wikipedia. Or read Thinking, Fast and Slow, by Daniel Kahneman, for even more insights into psychology you can apply to your marketing and persuasion efforts.
- A new book that is relevant is Using Behavioral Science in Marketing: Drive Customer Action and Loyalty by Prompting Instinctive Responses by MarTech contributor Nancy Harhut.
Getting back to ChatGPT, we know quite a bit about its strengths — it can be invaluable for helping you go from an empty page to a useful first draft — and its weaknesses, such as being susceptible to errors and bias as well as needing human work to make it fit with your brand and speak to your unique audiences.
But, even with these limitations, you can combine ChatGPT with effective persuasion tactics to become a true marketer — one who understands how humans work.
Using GenAI to write a prompt for a campaign brief
A key lesson to remember when working with ChatGPT, and indeed with any LLM (large language model) is that the results will be only as good as what you put into writing the prompt. When you layer specific persuasion tactics over your copy, your instructions must be even more specific.
Here’s the prompt I wrote for an email campaign which calls upon various persuasion principles and cognitive biases:
“You’re an email marketer for an ecommerce clothing retailer tasked with writing compelling copy to increase conversions for an email campaign. You’re selling a new range of summer dresses that are affordable, pretty, comfortable and stylish.
Deliver one subject line, opening paragraph, and call to action, using each of these persuasion principles and cognitive biases below, without actually mentioning the cognitive bias within the copy, but label each example with the principle or bias:
Reciprocity, scarcity, social proof, authority, anchoring, loss aversions, curiosity gap, endowment effect, confirmation bias, emotional appeal.”
Looks straightforward, right? But did you catch the eight specific instructions I included in the prompt? Let’s break it down:
1. Who the prompt is representing (the email marketer).
2. Whom the email represents (the ecommerce clothing retailer).
3. What I want it to do (write compelling copy to increase conversions).
4. What’s on offer (summer dresses).
5. Qualities of the product (affordable, pretty, comfortable and stylish).
6. The output (subject line, opening paragraph, and call to action).
7. The persuasion principles/cognitive biases (reciprocity, scarcity, social proof, authority, anchoring and framing, loss aversions, curiosity gap, endowment effect, confirmation bias, emotional appeal).
8. Special instructions (don’t mention the persuasion principle/cognitive bias in the copy).
These details are important to remember when working with these LLM tools. You will have to be quite specific, to keep it from either repeating your concepts verbatim in the results or from including incorrect or irrelevant material.
Most likely you’ll not get the results you want on the first pass. This is where you must call on your authority, knowledge and experience of persuasion marketing to know when ChatGPT gets it wrong. Take the time to refine to your prompt until what you see is close to what you want, reflects the persuasion principles you want to use, reflects your brand voice, and would appeal to your audience.
Tip: You can prime your session of ChatGPT to be ready to use these persuasion principles and cognitive biases, by asking ChatGPT to provide the definitions of the principles and biases you’re using.
How to choose a persuasion principle/cognitive bias
These principles are not interchangeable. Which one you pick will depend on what you know about your audience: what motivates them, what moves them to act, what their needs, interests or fears might be.
Guard against using the same principle over and over, on every campaign no matter what your goal might be for it.
Loss aversion, for example, is a popular one. We humans are hard-wired to avoid pain, such as a loss. It goes back to our caveman days, when we survived by running away from danger. Today, it’s still a popular motivation because we don’t want to miss a deadline. FOMO (Fear of Missing out) is real in today’s society.
But using it every time, especially on routine campaigns, can lead to the email equivalent of “banner blindness,” when website viewers got so used to banner ads that they learned to ignore them.
Social proof is used less often than it could be. Think of it as the human need to find the wisdom of the crowd, that we make decisions based on what others have done.
Testimonials, reviews, 5-star ratings, the number of people who have already purchased an item we feature in an email … all of these highlight social proof and can make us feel more confident in our decisions.
Anchoring. With this principle you are helping to set expectations for your customers. This is a common tactic in pricing. To start, you would feature a product or service at a mid-level or full price. Every price your customers will see will be anchored against that first price. You can then feature your discounted price next to it, and customers will measure that discount against the more expensive alternative.
Emotional appeal. As email marketers we always need to remember we are just people talking to other people and using words to elicit specific emotions that are relevant to the campaign.
This is what good copywriting is all about. We are using emotive words to connect our brand to our customers, to get their attention and persuade them to continue to buy from us.
Stay tuned for the results and testing
So, how did ChatGPT deliver on my prompt? You’ll have to wait for my next installment, where I reveal the first round of email drafts, where they succeeded and where they fell short. I’ll say this much for now – they illustrate why ChatGPT and other chat tools are best used as first drafts, not copy-and-paste output.
I will also explore how to use ChatGPT for testing in a scientific process that begins with a hypothesis and extends into strategic retesting to find the version that “wins” with your audience. And I’ll share some takeaways you can use as you build up your generative AI and testing strengths.
The post The perfect combination: GenAI and persuasion strategies for unbeatable A/B tests appeared first on MarTech.
MarTech(23)