Prompt integrity

19 Laws of ChatGPT Prompts

A playbook to keep your prompts sharp, grounded, and testable. Use these laws to avoid fluff, force trade-offs, and get responses you can act on.

LAW #1

Clarity > Cleverness

Avoid fancy phrasing. Say what you want in plain language.

ChatGPT rewards clarity, not poetry.

LAW #2

Start With the Friction

Begin your prompt with the thing you're stuck on.

That's the signal that matters most.

LAW #3

Prioritise Use Over Theory

Ask for advice like you're about to implement it.

E.g., "What would I do this week?" beats "Explain this conceptually".

LAW #4

Ground It In Reality

Reference something real: your site, customer, offer, calendar.

It forces relevant output.

LAW #5

Declare the Constraints

Don't just say what you want. Say what you don't want.

Avoid clichés, keep it under 75 words, mention [X].

LAW #6

Timebox the Lens

Set a time boundary: "Based on what’s true in June 2025…"

It frames everything.

LAW #7

Make It Choose

Present trade-offs.

E.g., "Which of these 3 ideas is most likely to convert?"

LAW #8

Ask It to Judge, Then Justify

Don’t just get output. Get critique.

E.g., "Rank these and explain your reasoning."

LAW #9

Simulate Real Scenarios

Use setups like:

"Act like this idea just flopped. What would you audit?"

"Pitch this to a skeptical investor."

LAW #10

Lock in the Persona

Tell it how to think, not just what to be.

E.g., "You are a calm, cutthroat operator who values speed over elegance."

LAW #11

Focus on Leverage

Instead of "give me ideas"

"What would be the highest-leverage move I’m overlooking?"

LAW #12

Make It Build With You

Prompt like you’re in a feedback loop:

“Here’s version 1. Make it 2x sharper.”

LAW #13

Include What You've Tried

Skip reinventing the wheel.

Already tested A/B/C. I’m stuck between D and E.

This reduces unnecessary re-prompts.

LAW #14

Push It Into Action

Wrap your prompt with:

"Now write the email/tweet/call script that would follow."

LAW #15

Give It Tone

Tell it how it should feel:

"Confident but not cocky."

"Smart and slightly confrontational."

Tone drives output.

LAW #16

Ask It to Steal Smartly

"Based on these 3 examples, write one that’s better."

This is leverage.

LAW #17

Use Negative Prompts

Tell it what to ignore.

E.g., "Avoid motivational language. Avoid filler phrases."

This helps you shape tone/style.

LAW #18

Invite Pushback

End with:

"What’s wrong with this approach?"

You’ll remove ChatGPT’s “yes-man” tendencies.

LAW #19

Refine Into a System

If it works once, make it reusable:

"Turn this into a prompt template I can run daily."