7 Common Mistakes When Using AI Prompts (And How to Fix Them)

1. Being Too Vague

(P) Be Precise & Specific. Don't just ask for "something about data." Be specific about the goal, format, length, and intended audience.

✅ Good Prompt: "Write a three-paragraph, engaging blog post about predictive maintenance for small business owners."

2. Asking Multiple Things at Once

(A) Address One Task Per Prompt. Avoid combining distinct tasks like summarising and drafting an email. AI performs better when focused.

✅ Good Prompt: "Summarise the main findings of this report and pull out five key action items."

3. Ignoring Role Context

(C) Give Context & Persona. The AI needs to know who it is writing for and who it is writing as to adjust its tone and technical level.

✅ Good Prompt: "Explain SQL joins as if I am a beginner data analyst who knows Excel, but is new to coding."

4. Forgetting to Provide Examples

(E) Provide Examples & Expectation. If you need a specific style, tone, or format, share a sample piece of content or describe the desired style in detail.

✅ Good Prompt: "Write a LinkedIn post about automation. Use a slightly provocative, conversational tone, similar to the sample post I shared above."

5. Not Iterating

(R) Refine, Reiterate, & Reroll. Don't expect the first answer to be perfect. Use the AI's output as a draft and treat the process as a conversation.

✅ Good Follow-up: "That summary is too formal. Rewrite it with a focus on highlighting the immediate business opportunity."

6. Overloading with Unnecessary Detail

Keep It Concise (Initially). Start with a simple, focused prompt. If the topic is complex, request a short overview first, then ask for expansions or details on specific sections.

✅ Good Prompt: "Provide a bulleted summary of the core principles of AI. Then, I will ask for more details on one principle."

7. Skipping a Final Review

Always Edit & Fact-Check. AI is fast, but it can 'hallucinate'. Outputs should always be treated as a first draft.

✅ The Rule: Never copy outputs straight into reports or public posts. Always verify names, numbers, dates, and tone before hitting send or publish.

Final Thoughts

Using the P.A.C.E.R. Method can help to create better prompts, which in turn will output better results:
Precise & Specific

• Address One Task

• Context & Persona

• Examples & Expectation

• Refine (Iterate)

Good prompts don’t need to be complicated. Be specific, give context, provide examples, and iterate. With practice, you’ll spend less time fixing AI outputs and more time using them to get ahead.

Shop Our best picks for productivity ➡️