The Best ChatGPT Prompts: 40 That Actually Work
A prompt you copy is worth more than one you admire. Forty of the best ChatGPT prompts across work, writing, code and study — each with the pattern behind it.
Type "write me an email" into ChatGPT and you get grey mush: correct, forgettable, the text nobody reads twice. The prompt was the problem, not the model. A good request hands ChatGPT a role, the context only you have, one clear task, and the exact shape you want back. Get those four right and the first draft is usually the one you keep. What follows is a curated shelf of the best ChatGPT prompts for real situations — writing, work, study, code, marketing, and the small daily decisions in between. Copy one, drop your details where the [BRACKETS] are, and bend it from there.
What makes a prompt one of the best
Skim any prompt below and you'll spot the same four moving parts. Learn them and every example becomes easy to adapt.
- Role — who ChatGPT should act as. "You are a hiring manager" pulls in the right vocabulary and judgement.
- Context — the background only you hold: the audience, the constraints, the source text.
- Task — one clear verb. Write, compare, plan, debug. A single sharp ask beats three fuzzy ones.
- Format — the shape of the answer: five bullets, a table, a 120-word email, valid JSON.
The best ChatGPT prompts aren't the longest. They're the ones where all four boxes are filled, and when a reply misses it's usually because one box was empty. Prefer to have the model build them? The ChatGPT prompt generator assembles one from a plain description, and the prompt optimizer tightens a draft you already have. These are ChatGPT-first; for the same idea across Claude and Gemini, see our AI prompt examples.
Writing and communication
A percentage target forces real cuts; "make it shorter" gets a polite trim.
Naming the emotion you want hidden controls tone better than "be professional."
Distinct angles get real variety, not ten rewordings of one line.
Scoping to three lines keeps feedback usable, not a wall of red ink.
Options side by side beat guessing the register you need.
Handing over your draft gets an edit, not a generic template.
Banning the clichés up front is faster than editing them out later.
Work and productivity
The fixed action-item shape drops straight into a tracker.
"Add nothing" stops the model padding your week with filler.
A reason per task exposes the busywork you were about to do.
"Not apologetic" cuts the over-sorry tone the model defaults to.
Rehearsing objections beats walking in with talking points alone.
Naming the constraints keeps the plan honest, not aspirational.
For a brief you reuse weekly, save it as a system prompt or in ChatGPT's Custom Instructions.
Learning and study
The misconception line surfaces the part you'd otherwise trip on.
Being tested reveals gaps that re-reading never does.
Asking for active recall gets practice, not a highlighting schedule.
One question at a time turns a list into a real study session.
Flagging load-bearing claims trains you to read critically.
In-line corrections keep the conversation flowing instead of stopping it.
A side-by-side table plus a hook beats re-reading the same two definitions.
If ChatGPT is still new to you, how to use ChatGPT covers the basics first.
Coding and tech
Capping it at three issues keeps the review actionable.
Cause-before-fix stops you pasting a patch you don't understand.
The edge-case ask turns a summary into something useful.
The false-positive example saves you a silent bug later.
"Without changing behavior" keeps a cleanup from becoming a rewrite.
A runnable snippet beats a paragraph of theory every time.
Marketing and business
"Outcome, not features" fixes the most common weak-copy habit.
Objections in hand, the rest of the funnel writes itself.
A different idea per post stops five near-identical summaries.
"No flattery" cuts the fake-personal opener everyone ignores.
Comparing angles beats committing to your first guess.
Splitting by angle gives you a real test, not six near-clones.
The gap is where your differentiation actually lives.
Everyday life and decisions
The grouped list turns a plan into something you can actually shop.
Questions first stops the model advising you blind.
Naming what you hate shapes the plan more than what you like.
A fallback ready keeps you from caving on the spot.
The "one clause" ask cuts through pages of boilerplate.
"Skip the obvious" gets you past the candle-and-mug defaults.
Diagnosing the failure first beats a generic 30-day plan.
How to make any prompt better
None of these are sacred. The quickest route to a better answer is usually a second message: tell ChatGPT what missed and it corrects fast. "Too formal, cut it in half" beats rewriting the prompt from scratch. A few habits lift almost any request:
- Show, don't just tell — paste an example of the output you want.
- Ask for the format up front, not after the first draft lands wrong.
- When a reply is vague, add the one constraint it ignored instead of starting over.
- Save the ones that work so you're not rebuilding them.
Keep your winners in a prompt library and reuse them instead of retyping. When a draft is close but not landing, run it through the prompt optimizer. The mechanics behind the best ChatGPT prompts — roles, constraints, Custom Instructions, memory — live in our ChatGPT prompts guide. Bookmark two or three of the ChatGPT prompts above, adapt them to your work, and you'll spend less time coaxing and more time using what comes back.
