Here’s a bad prompt: “Write blog posts about marketing.” » If someone gave you this instruction, would you know what they want and be able to deliver it the first time? No.

A better version would be “Create a three-point plan for a blog post on influencer marketing strategies for small e-commerce businesses in the Gen-Z fashion industry, focusing on Instagram and TikTok.” »

This prompt gives ChatGPT a clear roadmap, resulting in a more focused and valuable result.

So, when creating your prompt, include:

• Your target audiencelike local restaurants, tech industry professionals, etc.
• Specific objectives— for example, to increase brand awareness, to generate B2B leads, etc.
• All requirementslike “include real-life success stories” or “emphasize content marketing.”
• Preferred format-for example, a 500-word article or a five-step guide.
• Relevant general information— like “the brand focuses on sustainability”, “the guide must respond to technology industry practices”, etc.

Essentially, you’re giving ChatGPT a mini-brief, much like you would a human colleague.

2. Use role-play prompts

One of the best tricks to incentivize ChatGPT is to assign it a specific role or expertise. For example: “You are an experienced data scientist. Explain neural networks to a marketing team considering using AI in their marketing campaigns. Use analogies they will understand.

You’re more likely to get an explanation tailored to your audience’s background when you phrase your prompts this way.

You can ask ChatGPT to take on roles such as an industry expert, historical figure, or even a fictional character. Choose a role that matches the type of information or perspective you’re looking for, and the results can be great.

It’s also a great way to get new ideas. For example, you can ask ChatGPT to pretend it’s Don Draper, the fictional advertising executive from Mad Men. Ask him: “As Don Draper, come up with three ready-made content ideas to promote a new electric bike. » You might be surprised by the creative angles this approach can generate.

3. Show and tell in moments

Sometimes the best way to get what you want is to show ChatGPT an example. This technique, called a few-shot prompt, is great for generating consistent output.

Here’s an example prompt: “Generate 100 email addresses for fictitious marketers.” Follow this format:

1. [email protected]
2. …”

This technique works well for generating formatted content, creating content with a consistent style, or teaching ChatGPT to follow specific patterns or rules.

For more complex tasks, provide multiple examples. This effectively “trains” ChatGPT on the spot to produce the type of result you need in this case.

4. Take it step by step

Do you have an important and complicated request? Break it into bite-sized pieces. This approach gives you more control and often leads to better results.

For example, you might start with: “Let’s create a digital marketing plan for a new fitness app.” First, outline five key elements of a digital marketing strategy for mobile apps.

After you get this answer, continue with: “Great. Now, for each component, suggest two specific tactics suitable for launching a fitness app.

This method prevents ChatGPT from being overwhelmed or derailed, allows you to review and refine each step, and helps you identify areas that need improvement or expansion.

5. Use chain of thought prompts

Sometimes it’s good to understand how ChatGPT arrives at its answers, especially if it seems to be going off the rails or you’re unsure of the topic. In these cases, ask him to show his work!

Try something like: “You are a financial advisor creating a retirement savings plan for a 35-year-old who earns $80,000 a year. Walk me through your thought process step by step.

This technique gives you the answer while offering insight into the reasoning behind it.

You can improve this by asking follow-up questions at each step, such as “Why did you consider this factor?” » or “What alternatives have you considered here?” This creates a more interactive and educational experience and is great when using ChatGPT to learn about a new topic.

6. Compare and contrast for deeper insights

Ask ChatGPT to weighing options can give rise to nuanced and useful answers. This is perfect for decision-making scenarios.

To get even more benefit from this technique, specify the comparison criteriaask for a recommendation based on comparison or ask for a hybrid approach that combines the best elements of each option.

For example, you could use a prompt like “Compare the pros and cons of using influencer marketing versus paid social ads for launching a new skincare line.” Consider factors like cost, reach, and credibility.

7. Use hypothetical scenarios

Presenting imaginary situations to ChatGPT can stimulate your creativity and help you explore potential outcomesand provide interesting information.

To create effective hypothetical prompts:

• Start with a clear and imaginative premise.
• Specify the impact areas you want to explore.
• Encourage detailed and logical explanations.

For example: “Imagine a world where all marketing campaigns must be created using AI. Describe how this would change the advertising industry and consumer behavior, and explain why.

8. Flip the script with a reverse prompt

Sometimes, start with an answer and work backwards can unlock new ideas or deepen your understanding of a topic. ChatGPT can help you if you use reverse prompting.

Here’s an example: “The answer is “Our email sequence open rates increased by 50%.” » What helpful prompts or questions could a marketer have asked ChatGPT for guidance leading to this outcome? »

This technique is fantastic for brainstorming and idea generation. Start with a clear, specific outcome, ask for multiple paths that could lead to that outcome, and use follow-up questions to explore the most promising ideas.

9. Set Boundaries with Constraints Prompts

Addition specific limitations to your prompts can lead to more focused and creative responses.

Try a prompt like this: “Generate five effective prompts for LinkedIn posts on personal branding. Each prompt should be less than 200 characters and include a thought-provoking question.

Be clear and specific about boundaries, combine constraints to achieve more focused results, and experiment with different constraints to see how they affect the results.