Marketers are under constant pressure to create high-quality campaigns. But thanks to generative AI advancements like ChatGPT, they’re learning how to harness new technology to gain a tactical advantage.
Let’s be clear: ChatGPT and its large language model (LLM) competitors like Bing Chat, Google Bard and others aren’t creative services replacements. Like all generative AI tools, LLMs lack human emotion and understanding. They are, however, creative accelerators, bringing a wealth of practical possibilities to content ideation and research.
A recent Hubspot survey found 79% of marketers agree generative AI can improve content quality. With ChatGPT and other LLMs, marketers can tap into powerful capabilities to generate customized content, create high-level analysis of customer feedback and surface creative campaign ideas that captivate audiences.
With that in mind, we’ve outlined seven experiments below your marketing team can conduct with ChatGPT to maximize their creativity and time.
Four generative AI experiments for creating compelling content
When new employees join your team, what’s the first thing you do? You train them.
ChatGPT and other large language models (LLMs) are no different. In most cases, the AI can even learn your brand’s tone of voice and style. For example, enter some marketing copy or a blog post, and then prompt ChatGPT to “write a 500-word LinkedIn article about [a topic you provide] in the style of the above copy.” It will take some trial and error, but over time, you’ll be surprised at how accurately it aligns with your brand persona.
Here are four generative AI content experiments marketers should try:
- Outsource your campaign tactics brainstorm: You have the campaign brief, but what about the winning ideas? ChatGPT and internet-connected LLMs are ideal for early stage ideation when you have a general direction. Prompt the generative AI with your brief details and ask it to offer options.
- Automate your way to the perfect headline: The headline is the most important part of any piece, especially if you’re trying to sell or generate leads. Prompt ChatGPT or another LLM on the headline parameters you know will engage human readers (i.e. – listicle setups or queues like “how” and “why”), enter a synopsis of the article you’re about to write, and then ask it to create 10 compelling headlines. Don’t like those options? Run the prompt again. When you’ve found a headline you like, check it through a traditional search engine to make sure the LLM didn’t duplicate something that already exists.
- Create product descriptions: Need to nail the language about a new product? Try entering the specs and product use cases into an LLM and prompt it to create a brief synopsis with consumer-friendly, active language. At a minimum, this will give you a head start on your end output.
- Generate new angles that will engage your audience: Feel like you’re out of ideas? Input your target audience parameters and prompt ChatGPT to generate topics it believes will resonate with that group. Also, consider inputting constraints, such as content format or publishing platform.
Something to keep in mind in every LLM experiment is that the more detailed you make your prompts, the better the results you’ll likely receive. Be sure to use prompts that align with your campaign goals and messaging. Try prompts like “What if [unexpected event] happened?” or “How might we solve [specific problem]?” and don’t shy away from overexplaining your request.
“Generative AI should feel like a playground for marketers and creators alike,” said Eric Brandner, a content marketing director at Industry Dive’s studioID. “These applications are the ultimate blue sky space where you’re only limited by your curiosity. The key is to keep experimenting with new scenarios and prompts.”
Tips for creating the perfect ChatGPT prompt
- Clarity: Ensure that prompts are clear, concise and unambiguous.
- Specificity: Be specific about the task, format, or information required to guide any generative AI tool you use toward the desired output.
- Context: Provide sufficient contextual or background information to generate more relevant and accurate responses.
- Creativity: Experiment with various prompt structures, phrasings and formats to identify the most effective approach.
Three generative AI marketing research and analysis experiments
Imagine having market research panels and customer focus groups at your fingertips, any time of day or night. Better yet, how about the ability to create data-driven insights with a few keystrokes?
Just as with the creative process, LLMs aren’t a substitute for your research or analytics department, but they can be fed data to uncover insights and surface areas of opportunity. And they’re fast, too. ChatGPT can analyze large volumes of customer feedback data – such as online reviews, social media comments, and survey responses – and return high-level insights quickly. A brand manager could use ChatGPT to learn about product pain points, inform product development or marketing strategy, or generate customer personas and profiles based on demographic and behavioral data. (Though users should always remember to fact check any new data an LLM introduces, as ChatGPT’s “hallucinations” are a concern.)
With that in mind, here are three generative AI research and analysis experiments marketers should try:
- Anticipating pain points: Asking ChatGPT or another LLM about your customers’ biggest pain points, or how customers feel about products and features, can help you anticipate shifts in sentiment or other problems you may want to head off. Remember, your answers will only be as good as the data you enter, so come up with a variety of nuanced prompts and follow-up questions to see how the feedback changes.
- Conduct a messaging analysis: Paste your 10 most engaging social media posts or email blasts from the last year into an LLM and prompt it to analyze what they have in common. Again, the key is to be specific. Ask what sort of linguistic approaches they have in common. Ask if there’s a common emotional thread. And ask what could have been done better to see what new approaches it will suggest. An alternative – but perhaps even more useful approach – is to do this with posts or emails that underperformed your expectations, asking nuanced questions about areas where they may have fallen short.
- Pressure test your personas: Describe your key targets and ask ChatGPT or another LLM to create as many detailed personas as it feels are necessary. Prompting it to write follow–up descriptions after its initial insights may help you uncover new angles through which to view your audience. (This LinkedIn Learning course demonstrates how to do a basic persona creation exercise through ChatGPT.)
Human creativity is still king
It’s natural for creatives to be concerned about LLMs taking their jobs. But there’s no need to update your resume. ChatGPT and its competitors aren’t creative – they’re tools for creativity in the same way that Microsoft Office is a tool for writing, or Adobe Creative Cloud is a tool for design. Without people, those tools are useless.
Marketers need every advantage to stay ahead, and LLMs offer a wealth of tactical possibilities to drive engagement and deliver exceptional experiences. Think of generative AI tools as part of your team – the same as you’d use industry experts like those at Informa Tech to craft design compelling messaging speaks to your personas at different points of the customer journey – allowing you to lead strategically while saving significant time.
Finally, don’t forget that generative AI is always learning. As a best practice, be sure to provide regular feedback – good and bad – so your LLM of choice can sharpen its skills.
“Your creativity can shine through in your feedback,” said Liam Berry, a senior content strategist at Industry Dive’s studioID. How you correct and fine-tune the responses of the LLM will shape not just your current task, but its ability to assist in future projects.”
Informa Tech delivers high-performing digital marketing services across the customer journey. Learn how we can help you build your brand, earn trust, and drive demand.