The problem with generic prompts
You’ve probably seen them all over LinkedIn. The endless carousel posts and viral lists promising ChatGPT prompts that do something magical without needing to put in any work.
Sounds intriguing right? After all, all this AI stuff is so confusing, maybe this person has it all figured out. So you try them, only to get back generic garbage that falls flat, misses the point and doesn’t help you.
Let’s take a common generic prompt example floating around the internet:
“Hey ChatGPT, pretend you’re an expert researcher and write me a blog (or post or article) on [topic] for [audience]. Make it clear, add bullet points, and keep it under 1,000 words.”
Looks decent, right? But here’s why it falls flat:
- ChatGPT doesn’t know you, your voice, your brand or your target audience.
- It Doesn’t Know Your Intent. Are you trying to educate, persuade, or entertain? Sell a service? Drive newsletter signups?
Without that context generic prompts default to bland, boilerplate output and fills in the blanks with generic filler. Generic prompts, generic results.
What to do instead of generic prompts
If you want quality output, you need to ditch the generic prompts and start creating structured prompting that includes:
- Context: who you are, your audience, and what you’re trying to achieve.
- Voice and style: how you want it to sound.
- Clear instructions: formatting, length, and purpose.
- Examples: Show it what “good” looks like.
- Create clear instructions with specific intents and goals:
Generic prompts don’t work because AI tools work best when they receive specific instructions on what they need to do. Because they are predicting their responses word by word, the more specificity you can give them upfront, the better their outputs will be.
The main things to make sure you include in any prompt are intent (what are you trying to do) and goals (what a good output looks like). After, making sure the AI can get the general gist of your request, then you can start optimizing your prompt to make the outputs even better. - Create step-by-step requests to help let the AI “think”:
One of the best ways to getter high-quality results from AI is to give the AI time to “think.” The easiest way to accomplish this is to include “thinking” instructions in your prompt. For example, if you are asking the AI to solve a math problem, you should ask it to explain its work step by step.For more complex processes and workflows, you can even include a step-by-step process guide in your prompt and ask the AI to follow it to produce your output. For example, “You will be asked to extract important data from this article: First, save the name of the author. Then read the first paragraph to determine the primary topic of the article. Next, analyze the tone of the first paragraph to determine the sentiment of the article….”
- Use delimiters to direct the AI: Humans often use fonts, formatting, line breaks, and other visual hints to understand the significance of different parts of an article. To help AI understand the different input types from an instruction use “machine language” (a.k.a. delimiters) to mark breaks in the input formats.
For example, you may want to mark the end of user instructions and the start of examples for the AI. You can use symbols like ”’ “”” or <>
before and after your text to create distinct sections for your prompt. - Provide examples: The best way to train AI to understand what you want is to just show it what you want. If you can provide some examples of good outputs for a set of inputs in your prompt, your AI will do a much better job of predicting what you want the next time you put in a new input.
One thing to watch out for when providing examples is to ensure there is a varied set of inputs and outputs, so that the AI picks up on the right signals from your examples, rather than just producing the same output every time.
But let's take it a step further:
- Target Audience Doc: This tells an AI who you’re speaking to.
- Customer Avatar Doc: A representation of who your ideal customer is.
- Brand voice guidelines: How your (or your brand) presents itself, your business, targeted keywords, and so on.
- Examples of your writing: Any blog posts, articles, anything that you wrote that is at least 500 words and best represents your style.
Check out: Writing with AI: 5 Essential Tips to Keep Your Voice where I share the exact prompts that you can use to create your own branding and target audience docs that you can then use with any AI to guide your writing, research and other tasks that are specific to you and your business.
In conclusion
Bottom Line
- Generic prompts produce generic results.
- Structured, context-rich prompts produce content that sounds like you.
If you want to stand out and produce content that you can actually use, stop copy-pasting from generic prompt lists and start crafting prompts that accomplish specific tasks, reflect your brand, your audience, and your goals.
For more on building better prompts check out ChatGPT prompt examples – https://platform.openai.com/docs/examples
Multi-disciplinary IT support strategist. At SMB Consultants I specialize in bridging the gap between complex AI technologies with your actual business needs. Through 1:1 AI Coaching and Consulting my focus is on practical understanding, and implementation of solutions that solve a problem or address a specific business need.
