Last week I was at an AI conference and someone casually mentioned "just turn off the training data." The room nodded like this was common knowledge.
I realised I've been living in this space so long — using AI like a coworker, a strategist, a second brain — that I've completely forgotten most people have no idea what ChatGPT can actually do.
If that's you, first up: you're not behind. Despite what the LinkedIn bros will tell you, you're not. You're just playing catch-up in a space that's moving stupidly fast.
So I'm laying out the ChatGPT hacks that actually matter. Not the "optimise your workflow synergy" nonsense. Real, practical stuff that'll make your work faster, your brain calmer, and your business run smoother.
Let's go.
1. Turn off data training
This is non-negotiable.
If you're paying for ChatGPT Plus, you have a choice free users don't: whether OpenAI can use your conversations to train their model.
Where to find it:
- Go to your profile (bottom left)
- Click Settings
- Find "Data controls"
- Turn OFF "Improve the model for everyone"
That's it. Done. Your data stays yours.
If you're using the free version and you're worried about privacy, here's my litmus test: if you wouldn't post this on LinkedIn, don't drop it into ChatGPT.
That means financial data, client info, strategy that's close to your chest — all of that stays off the free version.
One more thing: that "Temporary Chat" button in the top right? It's not a privacy pass. Yes, conversations don't get saved to your memory. But OpenAI still keeps them for 30 days. If you work in an industry with strict data compliance, temporary chat is still not safe. Don't be fooled.
2. Custom instructions: the difference between mid and actually good output
This is where most people drop the ball. They either skip custom instructions entirely (wild), or they set them up so vaguely that ChatGPT has no idea what they actually want.
Think of it like onboarding a really smart high schooler with zero work experience.
They're brilliant. They're keen. But they have no context. That's your starting point.
What to include:
- How you think and write
- What you hate (words you never want to see again)
- What you use all the time (your lexicon, your jargon)
- Your tone rules
- What your business actually does
- Your values
Get stupidly specific about tone. Don't just say "casual" or "professional." That's vague enough that ChatGPT will default to the most generic version of that.
Setting it up:
- Go to Profile (bottom left)
- Click Personalisation
- Fill out the boxes:
- What you want ChatGPT to do — describe its role (e.g., "award-winning marketer with a decade of wins")
- How you want it to respond — your tone, your values, what matters
- About you — your business, the tools you use, your dream clients
- Capacity markers — "if I say low capacity, give me the simple version" or "if I say high capacity, push me with three options"
That last one is a game-changer. Some days you've got bandwidth and some days you don't.
3. The clarifying-questions hack: stop getting mid responses
You know that feeling when you ask ChatGPT something and it just... assumes? And gives you something that's fine but not really what you needed?
Here's the fix. At the bottom of your prompt, add:
"Can you ask me clarifying questions until you're 95% sure you can complete this task well?"
That forces ChatGPT to stop predicting and start asking. Instead of going off half-cocked with incomplete info, it'll ask you what it actually needs to know. Then you answer. Then it goes off and does the thing properly.
This is literally how you'd onboard a real human team member.
You can also use this to challenge yourself. Add:
"You're my sparring partner. Challenge me. Identify risks, blind spots, and opportunities I haven't thought of."
Otherwise it'll just agree with everything you say, which is comforting but useless.
4. The "act-as-top-0.1%" hack (my secret weapon)
This works in custom AI assistants and projects too. The idea is simple:
"Act as a top 0.1% marketer who specialises in [your thing]. You've worked for [relevant people/brands]. Complete this task."
Adding that constraint — top 0.1%, specific expertise, specific experience — instantly narrows the output from generic to specifically good. It's like creating a résumé for ChatGPT.
The more context you give it about who it's supposed to be and what calibre of thinking you want, the better the work.
5. Screenshot analysis: your personal data detective
Something people sleep on: you can just dump screenshots into ChatGPT.
- Need a Notion page reorganised? Screenshot it. Ask ChatGPT to restructure it.
- Collecting testimonials from Instagram DMs? Screenshot them all. Create a custom AI assistant that extracts the text so you can copy-paste to your website.
- To-do list chaotic? Screenshot it. Ask ChatGPT to prioritise by urgency and energy level.
- Screenshot your email marketing spreadsheet. Ask it to turn it into an interactive HTML dashboard. You can now filter stats and see what worked.
The more you give ChatGPT to see, the better it understands what you're asking.
6. Don't outsource your brilliance: edit, don't replace
Here's where most people mess up. They hand their work to ChatGPT and pretend it's done.
That's not using AI. That's outsourcing your thinking. And you'll lose your voice and the skills that built your business in the first place.
The better way: let AI help you articulate what you already know.
When you ask ChatGPT to review something, be specific:
- "Review this. Provide suggestions. Do NOT rewrite."
- "Rewrite this in my tone. Explain why you made those changes."
- "Can you make this clearer without changing my voice?"
Then you go back and forth. You push back. You disagree. You learn. That's when AI becomes useful instead of just another procrastination tool.
7. Voice mode: the hack that changes everything
We speak 3–5x faster than we type. If you're not using voice mode, you're leaving so much speed on the table.
- Open ChatGPT on your phone in voice mode (the sound wave icon)
- Brain dump everything you need to do. Ask ChatGPT to prioritise and help you figure out where to start
- Use a custom AI assistant on voice mode during your commute to extract content ideas
- Problem-solve in real time (we once fixed our washing machine by showing ChatGPT the hose)
- Create a "business mentor" custom AI assistant and chat with it when you're stuck on a decision
The key: you're not just chatting. You're sparring. And you're actually doing something with the ideas, not just consuming AI output forever.
Because yes — there's such a thing as consuming AI instead of creating with it. When you're stuck in loops of chatting and planning but never executing, you're just procrastinating.
ChatGPT will happily keep generating ideas for your coathanger business forever if you let it. Don't let it.
8. Automation: set it and forget it
You can set ChatGPT to run recurring tasks.
Connect your Gmail or Microsoft calendar and ask it to:
"Every Monday at 8am, send me a summary of everything I need to do this week based on my calendar."
Then it just... does it. Automatically.
You can also use this to find efficiency wins. Brain dump everything you did today (or a whole week) into voice mode. Ask ChatGPT: "What tasks are repeating? Which ones should I automate or delegate?"
It'll draw on your tech stack and suggest actual tools.
